Fish-Scale-Derived Type I Collagen as a Regenerative Biomaterial: Structure, Processing, Terahertz Characterization, and Ocular Applications within a Circular Bioeconomy

1. Introduction

Fish processing generates large volumes of by-products — skins, frames, heads, viscera and scales — a substantial proportion of which is discarded or diverted to low-value uses such as fishmeal despite containing structurally organised biopolymers and biominerals [1]. Global production has continued to expand, with aquaculture now supplying the majority of aquatic animals destined for human consumption, so the residue stream is both large and growing [2]. Interest in recovering value from these residues has developed alongside broader efforts to reduce material intensity in manufacturing and to close resource loops within food-processing supply chains [1].

Among the available residue streams, fish scales occupy a distinctive position. They are the only common fish-processing by-product that supplies both an organic protein fraction and a calcium phosphate mineral fraction from the same material [3,4]. Structurally, the scale is a mineralised biocomposite in which a Type I collagen matrix is reinforced by an apatitic phase closely related to bone mineral [4,5]. The collagenous component confers flexibility and biological recognition; the mineral phase provides stiffness and bioactivity. Because this organic–inorganic organisation broadly parallels that of vertebrate hard tissue, fish scales have been examined as a precursor for materials intended to interface with bone and soft tissue [4,5].

Relative to bovine and porcine feedstocks, fish-derived materials are associated with a lower perceived risk of transmitting mammalian pathogens and with fewer cultural and religious restrictions on use, and they are recovered from a residue stream that would otherwise require disposal [3,5]. These are legitimate advantages, but they concern hazard categories rather than product quality, and they do not substitute for the endotoxin, heavy-metal and microbiological control that aquatic feedstocks specifically require.

A second, quite separate line of interest connects fish biology to ophthalmology. Teleost fish retain a capacity for retinal regeneration that mammals have effectively lost: following injury, Müller glia in the fish retina dedifferentiate into proliferating progenitors that regenerate the lost neuron classes [38,39]. This has made the teleost retina a reference system for attempts to reactivate latent regenerative capacity in mammalian tissue. The two strands — fish scales as a source of implantable matrix, and fish retinal biology as a model of endogenous repair — are frequently discussed together because both address vision loss and both take their cue from fish. They are nonetheless independent lines of evidence, and a central aim of this review is to keep them clearly separated.

Despite an expanding literature on extraction and characterisation, the field remains concentrated at laboratory scale. Comparatively little published work addresses process scale-up, effluent management, cost structure, quality systems, or the specific circumstances of producer economies in which the raw material is abundant but the analytical and regulatory infrastructure is still developing. This asymmetry is the central problem addressed here: the constraint on industrial adoption is not primarily whether the target materials can be recovered, but whether they can be recovered reproducibly, cleanly, and to a specification that a purchaser will accept.

Accordingly, this review has four aims. First, it synthesises what is established about the structure of the scale and the principal recovery routes, distinguishing pathways supported by a consistent body of evidence from those that are not. Second, it examines characterisation, including the contribution and the limits of terahertz methods. Third, it assesses the ocular literature in detail and makes explicit the distinction between the material and conceptual contributions of fish biology. Fourth, it considers processing integration, industrial translation and the specific position of producer economies, separating conclusions supported by published data from strategic proposals advanced by the authors. The overall pathway from residue to application is shown in Figure 1.

2. Fish Scales as a Sustainable Biomaterial Feedstock

2.1. Hierarchical Structure and Composition

Fish scales are laminated structures in which an outer, heavily mineralised layer overlies a fibrillar collagenous base plate [4,6]. The organic phase is dominated by Type I collagen, which forms the structural framework of the scale [3,6]. The inorganic phase consists principally of calcium phosphate in an apatitic form, distributed through and around the collagen network rather than segregated as a discrete layer [4,7]. In one reported preparation retaining the native scale geometry, the mass fractions of water, organic matter and inorganic matter were 20.7%, 56.9% and 22.4% respectively [8], although such values are specific to species and preparation and should not be generalised.

The organisation of the base plate is a structurally interesting feature. Collagen fibrils are arranged in lamellae within which orientation is uniform, while successive lamellae are rotated relative to one another, producing a twisted-plywood or Bouligand arrangement [6,7]. The specific geometry is species-dependent: orthogonal plywood arrangements have been described in tilapia, twisted-plywood arrangements with interlamellar angles in the range of roughly 60–75° in several cyprinids and in Arapaima gigas, and double-twisted systems in other taxa [6]. In Arapaima, synchrotron small-angle X-ray scattering during tensile loading has shown that this arrangement governs the scale’s toughness through lamellar rotation, fibril straining and interfibrillar sliding [7]. This architecture is directly relevant to biomedical use, because ordered lamellar collagen is precisely the organisation required in a corneal stromal substitute, a point developed in Section 6.

This hierarchy is summarised in Figure 2, and the compositional and structural components of the scale are set out in Table 1. Compositional values reported for fish scales vary considerably between studies: species, age, habitat, diet and season all influence the collagen-to-mineral ratio and the calcium-to-phosphorus ratio of the mineral phase. This variability is frequently acknowledged in the primary literature but is rarely quantified systematically across species, and it is a material obstacle to specification-driven manufacturing rather than a minor experimental caveat.

The interpenetrating arrangement of the two phases carries a direct processing consequence. Because the mineral is distributed within the collagen matrix rather than layered separately, the separation chemistry chosen for one fraction necessarily determines what remains recoverable of the other. This constraint recurs throughout the review and is treated quantitatively in Section 8.

2.2. Recoverable Fractions and the Status of Evidence

Three recovery targets are discussed in the literature: Type I collagen from the organic matrix; hydroxyapatite and related calcium phosphates from the mineral phase; and, in a smaller number of reports, chitosan obtained by deacetylation of a chitinous residue said to be present in scales of certain species. Their characteristics, recovery routes and reported applications are compared in Table 1.

The evidentiary status of these three routes is not equivalent and treating them as equivalent is a recurring weakness in the applied literature. Collagen recovery is supported by numerous independent reports across multiple species and laboratories [3,8,10], as is hydroxyapatite recovery, principally by calcination but also by chemical routes [4,11]. Fish-scale-derived hydroxyapatite has been characterised as a candidate bone scaffold material [11], and whole decalcified scales have been evaluated as bioabsorbable fixation devices [12].

The chitosan route requires more careful handling than it usually receives. Chitin is the structural polysaccharide of arthropod cuticle, molluscan structures and fungal cell walls; it is not an established structural component of the vertebrate integument, in which the corresponding role is filled by collagen and keratin. Reports of chitosan recovered from fish scales are therefore not straightforwardly consistent with the accepted compositional picture. That said, the empirical literature is not negligible: chitosan has been isolated and characterised from the scales of Prochilodus magdalenae [13], from Labeo rohita [14], and from Nile perch alongside other locally available substrates [15], in each case with infrared, diffraction and thermal characterisation reported.

The appropriate reading is neither dismissal nor uncritical acceptance. What the published work establishes is that a deacetylated, chitosan-like polysaccharide fraction can be recovered from scale material by conventional demineralisation–deproteinisation–deacetylation sequences. What it does not yet establish, in the sources examined here, is the origin of that fraction, its mass balance against the starting material, or its structural identity by methods that discriminate chitin polymorphs from deacetylated protein residues sharing overlapping infrared bands. Confirmation would take the form of unambiguous structural identification — solid-state ¹³C NMR or diffraction consistent with chitin polymorphs — supported by elemental nitrogen analysis and a closed mass balance. Until then, the route should be described as reported but not independently confirmed at the level of origin, and industrial planning should not depend on it.

Set against other fisheries residues, the distinguishing feature of scales remains the co-presence of both principal fractions. Fish skin is an established and higher-yielding collagen source but carries no mineral phase to recover [3]. Crustacean shells are the conventional industrial source of chitin and chitosan but supply neither Type I collagen nor an apatitic mineral of biomedical interest. Bones and frames carry a substantial mineral fraction but a less accessible and less intact collagen fraction. Fish scales therefore permit a cascading recovery sequence from one feedstock, which is the principal argument for their industrial attractiveness. Whether that theoretical advantage survives contact with process economics is examined in Section 8.

3. Type I Collagen: Extraction, Processing and Properties

3.1. Structural and Biological Characteristics

Collagen is the most extensively investigated product recoverable from fish scales. The scale matrix is dominated by Type I collagen, the same genetic type that predominates in human skin, bone, tendon and cornea, and this correspondence underlies most of the interest in fish scales as a collagen source [3,8]. Reported properties include biocompatibility, enzymatic degradability, low immunogenicity and support for cell attachment and proliferation [3,10]. When extraction is conducted under controlled conditions, the native triple-helical conformation can be retained, which is the property that distinguishes intact collagen from gelatin and hydrolysates and determines whether the product is suitable for scaffold applications [3].

The property that most clearly differentiates fish collagen from mammalian collagen is its lower thermal stability, attributed to reduced imino acid content and correspondingly fewer stabilising interactions within the helix [3]. This relationship is not merely qualitative: across marine teleosts, denaturation temperature correlates with proline, hydroxyproline and serine content and tracks the physiological temperature of the source species [9]. The applied literature tends to present lower thermal stability either as a limitation or as an advantage depending on the argument being made, which is not a satisfactory treatment. It is more accurate to state that a lower denaturation temperature narrows the processing and sterilisation window and restricts use in load bearing or thermally demanding applications, while favouring applications in which rapid resorption or ready digestibility is desirable. Where thermal persistence is required, processing and crosslinking strategies can raise it substantially: a fish-scale collagen membrane prepared by a method that preserved the native scale geometry was reported to have a denaturation temperature of 79.5 °C [8]. Crosslinking in turn introduces its own regulatory considerations regarding residual crosslinker. The suitability of fish-scale collagen is therefore application-specific and should not be asserted in general terms. Table 3 sets out the reported characteristics of fish-scale-derived collagen relative to conventional mammalian sources.

3.2. Comparison of Extraction Approaches

Collagen recovery from scales follows a common sequence: cleaning and removal of non-collagenous protein, demineralisation, solubilisation, recovery by salting-out and dialysis, and drying [3]. The processing train and its control points are shown in Figure 3, and the principal solubilisation variants are compared in Table 2.

Table 2. Qualitative comparison of collagen extraction approaches reported for fish scales and related fish by-products. The table compares approaches on attributes discussed in the cited sources; it does not report or imply numerical yields, since the reviewed references do not permit a like-for-like numerical comparison.

Approach	Principle	Reported advantages	Reported or anticipated limitations	Main scale-up consideration
Acid-assisted extraction (typically acetic acid)	Solubilisation of the collagen network by disruption of intermolecular bonds under dilute acid [3]	Simple, widely reproduced, no specialised inputs; helix retained under controlled temperature [3]	Long contact times; high acid consumption; acidic effluent requiring neutralisation	Effluent volume and neutralisation cost dominate operating expenditure
Enzyme-assisted extraction (typically pepsin)	Proteolytic cleavage of telopeptide regions to release collagen [3]	Milder conditions; improved release from crosslinked matrix; shorter processing	Enzyme cost; risk of over-digestion and loss of helical integrity; requires an inactivation step	Enzyme cost and process-control capability
Alternative and green solvent systems (e.g. deep eutectic solvents)	Solubilisation in designed solvent systems intended to replace mineral or organic acids [3]	Lower reagent toxicity reported; potential for solvent recovery	Evidence base preliminary; solvent recovery and recycling not demonstrated in the reviewed sources 1	Solvent recycling determines both cost and the validity of the environmental claim
Decellularisation and decalcification with retention of native geometry	Removal of cellular material and chelation of the mineral phase without dissolving the collagen network [8,34]	Preserves the native lamellar organisation, which is the property of interest for corneal and barrier-membrane use	Product dimensions bounded by scale geometry; not applicable where a solubilised collagen feed is required	Reagent residues require toxicological justification; batch definition is per-scale rather than per-lot

¹ Solvent recovery is the determinant of whether the reduced-toxicity claim holds at production scale, and it is not addressed in the sources examined here. See Section 3.2.

Table 2. Qualitative comparison of collagen extraction approaches reported for fish scales and related fish by-products. The table compares approaches on attributes discussed in the cited sources; it does not report or imply numerical yields, since the reviewed references do not permit a like-for-like numerical comparison.

Approach	Principle	Reported advantages	Reported or anticipated limitations	Main scale-up consideration
Acid-assisted extraction (typically acetic acid)	Solubilisation of the collagen network by disruption of intermolecular bonds under dilute acid [3]	Simple, widely reproduced, no specialised inputs; helix retained under controlled temperature [3]	Long contact times; high acid consumption; acidic effluent requiring neutralisation	Effluent volume and neutralisation cost dominate operating expenditure
Enzyme-assisted extraction (typically pepsin)	Proteolytic cleavage of telopeptide regions to release collagen [3]	Milder conditions; improved release from crosslinked matrix; shorter processing	Enzyme cost; risk of over-digestion and loss of helical integrity; requires an inactivation step	Enzyme cost and process-control capability
Alternative and green solvent systems (e.g. deep eutectic solvents)	Solubilisation in designed solvent systems intended to replace mineral or organic acids [3]	Lower reagent toxicity reported; potential for solvent recovery	Evidence base preliminary; solvent recovery and recycling not demonstrated in the reviewed sources 1	Solvent recycling determines both cost and the validity of the environmental claim
Decellularisation and decalcification with retention of native geometry	Removal of cellular material and chelation of the mineral phase without dissolving the collagen network [8,34]	Preserves the native lamellar organisation, which is the property of interest for corneal and barrier-membrane use	Product dimensions bounded by scale geometry; not applicable where a solubilised collagen feed is required	Reagent residues require toxicological justification; batch definition is per-scale rather than per-lot

¹ Solvent recovery is the determinant of whether the reduced-toxicity claim holds at production scale, and it is not addressed in the sources examined here. See Section 3.2.

Table 3. Reported characteristics of fish-scale-derived Type I collagen relative to conventional mammalian sources. Entries are qualitative statements of what the cited literature reports; no numerical performance comparison is asserted, since the reviewed sources do not provide like-for-like measurements.

Attribute	Fish-scale-derived collagen	Conventional bovine or porcine collagen	Consequence for use
Genetic type	Predominantly Type I [3,8]	Predominantly Type I	Correspondence with human skin, bone, tendon and cornea supports substitution in principle
Thermal stability	Lower denaturation temperature; correlates with imino acid content and source-species physiological temperature [3,9]	Higher denaturation temperature	Narrows the processing and sterilisation window; favours rapidly resorbing applications; crosslinking or geometry-preserving processing can raise it substantially [8]
Zoonotic risk profile	No mammalian transmissible spongiform encephalopathy risk [3]	Recognised hazard category requiring source control	A hazard-category advantage, not a product-quality advantage; aquatic-specific contaminant control remains necessary
Cultural and religious acceptability	Fewer restrictions on use [3,8]	Restrictions apply in several contexts	Widens the addressable market in defined segments
Native structural organisation	Base plate already organised into oriented lamellae [6,7]	Organisation depends on the source tissue and is usually lost on solubilisation	Directly relevant to corneal and barrier-membrane applications, where lamellar order must otherwise be engineered
Feedstock status	Residue stream, currently unpriced [1]	Established commercial supply chain	Low nominal input cost, offset by collection, stabilisation and quality-assurance costs
Dimensional constraint	Product dimensions bounded by scale geometry where native architecture is retained	Not dimensionally constrained in the same way	Limits large-area implants; favours tectonic and lamellar indications

Table 3. Reported characteristics of fish-scale-derived Type I collagen relative to conventional mammalian sources. Entries are qualitative statements of what the cited literature reports; no numerical performance comparison is asserted, since the reviewed sources do not provide like-for-like measurements.

Attribute	Fish-scale-derived collagen	Conventional bovine or porcine collagen	Consequence for use
Genetic type	Predominantly Type I [3,8]	Predominantly Type I	Correspondence with human skin, bone, tendon and cornea supports substitution in principle
Thermal stability	Lower denaturation temperature; correlates with imino acid content and source-species physiological temperature [3,9]	Higher denaturation temperature	Narrows the processing and sterilisation window; favours rapidly resorbing applications; crosslinking or geometry-preserving processing can raise it substantially [8]
Zoonotic risk profile	No mammalian transmissible spongiform encephalopathy risk [3]	Recognised hazard category requiring source control	A hazard-category advantage, not a product-quality advantage; aquatic-specific contaminant control remains necessary
Cultural and religious acceptability	Fewer restrictions on use [3,8]	Restrictions apply in several contexts	Widens the addressable market in defined segments
Native structural organisation	Base plate already organised into oriented lamellae [6,7]	Organisation depends on the source tissue and is usually lost on solubilisation	Directly relevant to corneal and barrier-membrane applications, where lamellar order must otherwise be engineered
Feedstock status	Residue stream, currently unpriced [1]	Established commercial supply chain	Low nominal input cost, offset by collection, stabilisation and quality-assurance costs
Dimensional constraint	Product dimensions bounded by scale geometry where native architecture is retained	Not dimensionally constrained in the same way	Limits large-area implants; favours tectonic and lamellar indications

Acid-assisted extraction, typically with acetic acid, remains the reference method because it is simple, requires no specialised inputs and is well described [3]. Its disadvantages are long processing times, substantial acid consumption and an acidic effluent that must be neutralised. Enzyme-assisted extraction, usually with pepsin, cleaves telopeptide regions, shortens processing and can improve recovery from crosslinked matrices; the constraints are enzyme cost, the need for an inactivation step, and the requirement to control proteolysis so that the helix is not degraded [3]. Alternative solvent systems, including deep eutectic solvents, have been proposed as lower-toxicity replacements for conventional acid treatment and have been compared directly with acetic acid extraction [3]. The evidence for these systems remains preliminary in respect of the property that determines whether their environmental claim holds at production scale, namely solvent recovery and recycling; this is not demonstrated in the sources examined here, and the sustainability claim should be regarded as provisional rather than established.

Demineralisation deserves separate emphasis because it is the step that most distinguishes scales from skin as a collagen feedstock. The high calcium content of the scale, present as apatite, must be removed before the collagen network becomes accessible, and chelation with EDTA is the commonly used approach [3]. The completeness of this step conditions everything downstream: residual mineral compromises optical clarity in ocular applications and confounds the mass-balance basis on which yield is reported.

A general observation applies across all these variants. Comparisons between published extraction studies are impeded by inconsistent reporting. Yield may be expressed on a wet, dry or defatted basis; purity may be assessed by different criteria; and helix integrity is not always verified at all. This heterogeneity, rather than any deficiency of the methods themselves, is the main reason that no ranking of extraction approaches by performance can responsibly be given here.

3.3. Purification, Fabrication and Sterilisation

Recovery is followed by salting-out, dialysis against dilute acid or buffer, and drying, most often by lyophilisation. Each step carries a specific risk: incomplete dialysis leaves residual salt that alters subsequent gelation; uncontrolled freezing during lyophilisation determines pore geometry in the dried construct and therefore its behaviour as a scaffold.

Fabrication routes reported for fish-derived collagen include casting into films and membranes, freeze-drying into porous sponges, directional freezing to impose anisotropic pore alignment, electrospinning, and extrusion-based bioprinting [5]. For applications requiring mechanical persistence or resistance to enzymatic degradation in vivo, crosslinking is standard; carbodiimide chemistry (EDC, usually with NHS) is the most frequently reported approach in ocular work, with genipin and dehydrothermal treatment as alternatives.

Terminal sterilisation is an underexamined step in this literature and interacts directly with the thermal-stability constraint discussed above. Moist-heat sterilisation is generally incompatible with an unmodified fish collagen construct; the alternatives — gamma or electron-beam irradiation, ethylene oxide, and supercritical carbon dioxide — each carry the potential to alter crosslink density, degradation rate or residual-chemical burden. Any construct intended for implantation must be characterised after sterilisation rather than before it, and the reviewed literature does not consistently do so.

4. Characterisation of Fish-Scale-Derived Collagen

4.1. The Conventional Analytical Toolbox

No single technique establishes that a recovered material is intact Type I collagen of adequate purity; the assessment is necessarily convergent, and the principal methods and the information each yields are summarised in Table 4 and Figure 4.

Infrared spectroscopy is the workhorse. The amide bands of the polypeptide backbone provide both identification and conformational information: amide A near 3300 cm⁻¹ (N–H stretching), amide B near 2920 cm⁻¹ (C–H stretching), amide I near 1650 cm⁻¹ (predominantly C=O stretching, and the band most used for secondary-structure assessment), amide II near 1550 cm⁻¹ (N–H bending combined with C–N stretching), and amide III near 1240 cm⁻¹ (C–N stretching and N–H bending) [19,20,21]. The ratio of the amide III band to the 1450 cm⁻¹ band is conventionally used as an index of triple-helical integrity, and infrared spectroscopy has been applied specifically to discriminate between collagen types [19]. Lower-frequency bands, sometimes designated amide IV to VI and lying below roughly 750 cm⁻¹, report skeletal vibrations of the backbone and out-of-plane bending modes; these have been characterised in model amide systems by far-infrared spectroscopy [22] and carry information on overall conformation and intermolecular interaction rather than on secondary structure directly.

Electrophoresis under denaturing conditions resolves the α1 and α2 chain pattern and supports assignment to collagen type I. Ultraviolet–visible spectroscopy shows the characteristic peptide-bond absorption in the 210–240 nm region, attributable to C=O, carboxyl and amide functionalities in the polypeptide chains. X-ray diffraction identifies the apatite phase and reports crystallinity in mineral-containing preparations. Scanning electron microscopy with energy-dispersive analysis reports pore architecture and the calcium-to-phosphorus ratio. Differential scanning calorimetry gives the denaturation temperature and is the direct measure of the thermal-stability constraint discussed in Section 3.1 [9], while thermogravimetric analysis resolves the organic-to-mineral mass balance. For ocular applications specifically, transmittance, haze and refractive index measurements are required and are discussed in Section 6.

4.2. Terahertz (T-Ray) Spectroscopy and Imaging

The terahertz region, spanning roughly 0.1 to 30 THz, probes low-energy excitations: collective vibrational modes of macromolecules, hydrogen-bond network dynamics, and librational and torsional motions that are inaccessible to mid-infrared methods sensitive to localised bond vibrations. Terahertz spectroscopy has been applied to collagen since the earliest surveys of biomolecular response in this range, in which the absorbance of collagen, DNA and serum albumin was compared between 0.1 and 2.0 THz [23]; subsequent work has developed the analysis of peptide and protein spectra in this region [24] and its application to biomolecule detection more broadly [27].

Three attributes make the approach attractive for biomaterial characterisation. It is label-free and non-destructive, so the same specimen can be measured and then used. It is strongly sensitive to water, since the dielectric response in this range is dominated by hydrogen-bonded network dynamics, which makes it a candidate probe of hydration state and, by extension, of drying and rehydration processes [26]. And terahertz radiation penetrates many non-conductive, optically opaque materials, permitting sub-surface imaging of structures that visible-light microscopy cannot reach [25,26].

Recent reviews position terahertz spectroscopy and imaging as emerging tools in tissue engineering specifically, for evaluating the mechanical and hydration characteristics of scaffolds non-invasively, while noting that limited penetration depth and system cost currently restrict wider adoption [26]. Continuous-wave terahertz imaging has been demonstrated on biomedical specimens [25]. Instrumentation based on dendrimer dipole excitation has been developed as a compact source and detection route for spectrometry and molecular imaging in this range [28,29].

4.3. An Illustrative Case: T-Ray Imaging and Spectra of a Rohu Scale

To indicate concretely what terahertz measurement of a fish scale yields, illustrative data acquired by one of the present authors (A.R.) on a scale of rohu (Labeo rohita) are reproduced in Figure 5 and Figure 6, extending an earlier account of terahertz approaches to fish-scale valorisation [30]. These are presented as a demonstration of the class of information the technique provides, not as a validated characterisation of a biomaterial product; they have not been subjected to replicate measurement, inter-specimen comparison or cross-validation against the conventional methods of Section 4.1, and no conclusion in this review rests on them.

The imaging modality is a cameraless raster-scanning arrangement rather than a detector array: a single source–detector pair samples the specimen point by point across a grid, and the resulting dataset is reconstructed into two- or three-dimensional images [28,29]. Figure 5 shows the mounted specimen and a nested series of surface fields from approximately 5 mm × 5 mm down to 200 µm × 200 µm, together with a reconstructed volume of 10 × 10 × 8 µm. The concentric ridge pattern characteristic of cyprinid scale growth is resolved at the coarser fields, and progressively finer surface texture at the higher magnifications.

Figure 6a shows a T-ray absorbance spectrum from the same specimen, computed as a Lomb periodogram — an estimator appropriate to unevenly sampled data [29] — over approximately 0.1 to 32 THz. The spectrum is broadband with numerous overlapping features, which is what a heterogeneous biological composite containing protein, mineral and water would be expected to produce. Figure 6b shows an eigen-frequency estimate derived from the same data, in which a sharp rise at low frequency is followed by a plateau at approximately −130 dB extending across the remainder of the range.

Interpretation of these features requires care, and the distinction between what is observed and what is inferred should be maintained. In the raw spectrum, features in the low-frequency region are plausibly associated with collective motions of the collagenous matrix, including inter-chain hydrogen bonding, since collective protein modes are known to fall in this region [23,24]. Features at higher frequency may reflect lattice vibrations of the apatite phase or coupling between mineral and organic matrix. These are proposed assignments consistent with the composition of the specimen; they are not established assignments, because band assignment in this frequency range is model-dependent and, in a multicomponent hydrated sample, individual modes are broadened and overlapped to a degree that precludes unique attribution without corroborating measurement.

The derived spectrum in Figure 6b should be read conservatively. Its dominant characteristic is the low-frequency rise followed by a broad plateau; the plateau most plausibly reflects bulk density and scattering rather than discrete resonances, and the trace does not display well-separated peaks against that background. Claims that specific eigen-frequencies constitute confirmed molecular signatures of collagen or hydroxyapatite in this specimen are therefore not supported by the data as presented and are not made here.

4.4. What Terahertz Methods Can and Cannot Establish

Because terahertz characterisation is sometimes presented in the applied literature as a validation method in its own right, its scope should be stated precisely.

Terahertz methods can plausibly contribute non-destructive detection of gross structural or compositional change between specimens; sensitivity to hydration state, which is relevant to monitoring drying, rehydration and possibly the progress of demineralisation; sub-surface imaging of internal architecture without sectioning; and, in principle, a rapid inline screening signal in a manufacturing setting, where the requirement is to detect deviation from a reference rather than to identify composition ab initio.

Terahertz methods, on the present evidence, cannot substitute for mid-infrared spectroscopy in confirming triple-helical integrity, for diffraction in identifying mineral phase, for calorimetry in determining denaturation temperature, or for electrophoresis in assigning collagen type. Nor does an absorbance spectrum report mechanical properties: absorbance and modulus are not related by any established relation in this system, and inferences from one to the other are unwarranted. The realistic role is complementary — a fast, non-destructive comparator operating alongside, and calibrated against, the established methods summarised in Table 4, not a replacement for them. Realising even that role would require demonstration of measurement reproducibility across specimens and instruments, establishment of a reference spectrum for a material of known specification, and determination of the sensitivity of the response to the process variables of interest. None of these has been reported for fish-scale collagen.

5. Fish-Scale Collagen in Regenerative Medicine

Beyond ophthalmology, fish-derived collagen has been examined across a range of regenerative contexts, and the breadth of that work provides the background against which the ocular evidence should be read. Reported applications are summarised in Table 5.

The general case for fish collagen as a scaffold material rests on its capacity to provide a porous, cell-adhesive substrate with a degradation profile compatible with tissue ingrowth [10]. Advances in scaffold fabrication have extended the range of architectures achievable from fish-derived matrices, and recent reviews position these materials as a response to donor-tissue scarcity across several tissue types [5].

In bone and periodontal applications, the native composite architecture of the scale is itself the point of interest, since the organic–inorganic organisation approximates that of bone. Fish-scale-derived hydroxyapatite has been prepared and characterised as a candidate bone scaffold material [11], and collagen membranes prepared from scales with retention of the native geometry have been evaluated as barrier membranes for guided bone regeneration, with osteogenic differentiation of bone-marrow mesenchymal stem cells assessed in vitro [8]. Decalcified whole scales have been investigated as bioabsorbable fixation devices for fracture [12]. In soft-tissue repair, acellular fish-scale scaffolds loaded with mesenchymal stem cells have been applied to skin flap regeneration [18], and nanoparticles fabricated from tilapia-scale collagen have been evaluated for wound healing, with the retained bioactivity supporting cellular attachment [17].

Two qualifications apply across this literature. First, the great majority of reported work terminates at material characterisation and in vitro assessment, with a smaller body of small-animal work and very little beyond it; the stage of evidence, not merely the existence of a report, is what should govern interpretation, and Table 5 is annotated accordingly. Second, the domains differ sharply in the evidence and regulatory dossier they require. Cosmetic and nutraceutical uses of collagen and its hydrolysates are commercially established and are constrained mainly by purity and specification consistency; wound-contact and implantable applications require biocompatibility testing and quality-system compliance. Conflating these domains overstates the maturity of the field and claims of clinical efficacy for ingested collagen peptides are contested in the wider literature and are not evaluated here.

6. Ocular Regenerative Medicine: Fish-Scale Collagen and the Biocornea

6.1. The Clinical Problem and the Structural Rationale

Corneal opacity is a leading cause of blindness and of moderate-to-severe visual impairment, and the shortage of human donor tissue is severe: recent analysis places the availability at roughly one donor cornea for every seventy patients in need [36]. That gap has driven sustained work on bioengineered alternatives, of which biosynthetic collagen-based implants have reached clinical evaluation — a recombinant human collagen implant was assessed in a phase 1 study with twenty-four-month follow-up, establishing that a collagen scaffold can support host corneal regeneration in humans [37].

The structural rationale for fish scales in this context is specific and worth stating precisely. Corneal transparency and mechanical strength depend on an ordered extracellular matrix of Type I collagen in which fibrils of small and uniform diameter are densely and regularly arranged; disruption of that order produces opacity. The collagenous base plate of the fish scale is not merely a source of Type I collagen but is already organised into lamellae with defined interlamellar orientation [6,7]. It is this pre-existing lamellar order, rather than the collagen content alone, that distinguishes scale-derived matrices from collagen reconstituted from solution, in which comparable order must be imposed by directional freezing, moulding or printing. The pathway from scale to implant is shown in Figure 7.

6.2. Processing Route

The reported route to a corneal matrix differs from the solubilisation route of Section 3 in that the scale architecture is deliberately retained rather than dissolved. Scales — most often from tilapia (Oreochromis spp.), with grass carp also reported — are harvested from processing waste, cleaned, and then decellularised and decalcified by sequential treatment, typically involving sodium hydroxide, phosphate-buffered saline and sodium chloride solutions for cellular removal and EDTA for chelation of the mineral phase [34]. The result is a soft, transparent collagen lamellar matrix. Implants are then trimmed to defined diameter, in reported work by laser punch [34], and may be crosslinked, commonly with carbodiimide chemistry, to improve mechanical properties and resistance to enzymatic degradation in vivo.

6.3. What the Evidence Base Actually Shows

The ocular literature on fish-scale matrices is unusually well developed for a material of this kind, and it is worth setting out by stage rather than in aggregate. Evidence levels are compared in Table 6.

At the level of material properties, a collagen lamellar matrix prepared from grass carp scales was characterised for transparency, refractive index, mechanical properties and cellular compatibility, and judged a plausible candidate for corneal tissue engineering [35]. Light scatter and transmission of a fish-scale-derived collagen matrix have been compared directly with human cornea [34].

At the level of cytocompatibility, a fish-scale-derived collagen matrix was assessed against primary human corneal epithelial and stromal cells obtained from donor corneas, examining direct and indirect cytotoxicity, proliferation, corneal cell phenotype and adhesion markers [32]. Earlier work had established that the scaffold supports repopulation by rabbit corneal stromal cells and has high oxygen permeability [31].

At the level of in vivo evaluation, the material has been examined in a graded sequence of models. Pocket implantation in rats established safety, biocompatibility and regenerative potential, together with an assessment of properties relevant to use as an artificial cornea [31]; in vivo immunogenicity was subsequently characterised [32]. In rabbits, implants were followed for six months, with corneas reported clear at three and six months, no haze or ulceration observed, filling of the interface between implant and host stroma, and no observable immune-cell infiltrate at day 180; the implant remained transparent without apparent fragmentation [33]. In minipigs, the material was used to seal 2 mm full-thickness corneal perforations, with follow-up over several days assessed by slit lamp and optical coherence tomography and confirmed histologically [34].

The stage that has not been reached is equally important. No controlled human clinical trial of a fish-scale-derived corneal implant is identified in the sources examined here. The porcine work was explicitly a short-duration test of tectonic sealing rather than of long-term integration or optical outcome [34], and the authors themselves noted that clinical cases of the specific injury modelled are uncommon. Comparative systematic review of bioengineered corneal grafts in animal models places this class of material within a broader field in which several scaffold strategies remain at preclinical stage [36]. The appropriate summary is that fish-scale-derived collagen matrices have accumulated a consistent and reasonably deep preclinical record across four species, and that this is a stronger position than most candidate corneal biomaterials occupy, but that it is not clinical evidence.

6.4. Constraints Specific to Ocular Use

Four constraints recur and are set out in Figure 7. Geometry is the first and is intrinsic to the feedstock. A scale is a small, curved object of fixed dimensions, so implant diameter and thickness are bounded by the size of available scales; large-diameter grafts cannot be obtained from a single scale, and the curvature of the native scale must be reconciled with corneal curvature. This is a genuine limitation on the range of clinical indications the material can address, and it points toward tectonic and lamellar applications rather than full-thickness replacement.

Thermal stability is the second, and it interacts with sterilisation as discussed in Section 3.3. The processing window is narrow, and any terminal sterilisation step must be shown not to compromise transparency or mechanical performance.

Residual reagents are the third. Decellularisation and decalcification employ hydroxide, chelating and salt solutions, and crosslinking introduces carbodiimide or equivalent chemistry; residues of each require toxicological justification under device regulation, particularly for an implant in contact with the ocular surface.

Variability is the fourth and connects back to Section 2.1. Species, season and farming conditions alter scale composition and lamellar organisation, and therefore transparency and mechanics. Farming tilapia under controlled conditions has been noted as a route to consistency in scale supply, but no accepted release specification for a scale-derived corneal matrix exists, and this — rather than any deficiency in the underlying biology — is the most immediate barrier to clinical translation.

7. Fish Retinal Regeneration as a Biological Model

7.1. The Distinction That Must Be Maintained

The second contribution of fish biology to ophthalmology is conceptual rather than material, and the distinction is made explicit in Figure 8. Fish scales supply a substance that is processed and implanted. Teleost retinal regeneration supplies a mechanism that researchers hope to reproduce in host tissue. No fish-derived material is implanted in the second case, and fish-scale collagen has no demonstrated capacity to induce retinal neurogenesis. The two operate on different segments of the eye, carry different regulatory classifications, and have different failure modes and translational timelines. They are discussed together in this review because both derive from teleost biology and both address vision loss, but treating them as a single programme overstates the maturity of both.

7.2. The Mechanism in Teleost Fish

Müller glia are the principal glial component of the retina, maintaining retinal homeostasis and providing structural support. In mammals, their response to injury is variable and can be protective or detrimental; although they can be induced to proliferate and generate neurons under particular conditions, these responses are insufficient to repair a damaged retina. In teleost fish such as zebrafish, by contrast, injury triggers a reprogramming event that confers retinal stem-cell characteristics, producing a proliferating progenitor population capable of regenerating all major retinal cell types and restoring vision [38].

Mechanistically, the sequence involves dedifferentiation, interkinetic nuclear migration, asymmetric division and subsequent differentiation of progenitors into the lost neuron classes. Growth factors, Wnt ligands and cytokines converge on MAPK/ERK, GSK3β/β-catenin and JAK/STAT signalling to drive injury-dependent expression of ascl1a, a basic helix–loop–helix transcription factor that regulates most aspects of the regenerative programme, including downstream lin28 and let-7 microRNA signalling [38]. Comparative work across vertebrate model systems has established both the conservation and the divergence of these pathways [39], and the zebrafish has become a standard model for ocular translational research spanning repair and regeneration [43].

7.3. Translation to the Mammalian Retina

The translational question is why mammalian Müller glia do not do this and whether they can be induced to. The evidence points to active suppression rather than simple absence of capacity. Ascl1 is not upregulated in adult mouse retina following excitotoxic damage, whereas exogenous expression can reprogramme mammalian Müller glia toward neuronal fates; combining Ascl1 overexpression with histone deacetylase inhibition stimulated functional neuronal regeneration from Müller glia in adult mice [40]. Signalling that is pro-regenerative in zebrafish can be inhibitory in mammals — STAT signalling is a documented example of this species divergence — and microglial activation and other features of the mammalian retinal microenvironment further constrain reprogramming [41]. Recent work has identified intercellular transfer of Prox1 as a mechanism suppressing regenerative potential, with its disruption restoring some capacity [42].

Two cautions are warranted. First, the great majority of this work is in rodent and zebrafish models; no clinical therapy based on Müller glial reprogramming is established, and the efficiency, fidelity and functional integration of reprogrammed cells remain active questions. Second, the specific signalling claims in this area are numerous and not uniformly replicated across models and injury paradigms, so individual mechanistic assertions should be attributed to their model system rather than generalised to the vertebrate retina.

8. Integrated Processing, Scale-Up and Biorefinery Concepts

The argument that fish scales are industrially attractive rests on recovering more than one fraction from a single feedstock. This section examines that argument on its merits, because it is frequently asserted and rarely tested. A cascading arrangement is shown in Figure 9a: controlled demineralisation liberates a calcium-rich liquor from which calcium phosphate can be precipitated, the demineralised residue is solubilised to recover collagen, and the insoluble remainder may be subjected to alkaline treatment.

Three constraints limit how far this concept can be pushed, and they are structural rather than incidental.

First, the separation chemistries compete. Conditions that maximise mineral recovery, particularly direct calcination, destroy the collagen entirely [4]; conditions that preserve collagen integrity require mild demineralisation, which recovers the mineral in dissolved form requiring further processing into a usable solid [3]. A cascade is therefore not a free addition of revenue streams but a compromise in which each fraction is recovered at less than the yield or quality achievable if it were the sole target. The choice between a mineral-first and a collagen-first route is made at the separation step and cannot be deferred.

Second, each additional recovery step adds capital cost, reagent consumption, energy demand and effluent volume. The environmental case for valorisation depends on these burdens remaining smaller than those of the disposal and virgin-material production they displace, which is an empirical question. No life-cycle assessment of fish-scale valorisation was identified in the sources examined here, and in its absence the environmental claim is an assumption rather than a conclusion.

Third, the economics of a multi-product plant are set by its highest-value product, not by the total mass recovered. Recovering a low-value fraction at the expense of the quality of a high-value fraction reduces plant revenue. Integration is therefore justified where the secondary fraction is genuinely additive and where its recovery does not compromise the primary product, and not as a matter of principle.

These constraints do not defeat the integrated concept, but they mean that the design question is which subset of fractions to recover, in what order and at what quality target. Answering it requires process data — closed mass balances at pilot scale, reagent and energy inventories, effluent characterisation — that the current literature does not supply. The barriers and the corresponding technical requirements are set out in Table 7.

9. Industrial Translation in a Producer Economy: The Case of Bangladesh

The generic barriers described above take a particular form in economies where the feedstock is abundant but the supporting infrastructure is not. Bangladesh is a useful test case, and the discussion below separates statements supported by the reviewed literature and official statistics from strategic proposals advanced by the authors; the latter are identified as such and should not be read as findings.

9.1. Feedstock Availability and Logistics

Supported by data. Bangladesh is among the world’s leading fish-producing countries. Total production reached 5.018 million metric tonnes in fiscal year 2023–24, of which inland closed-water aquaculture contributed 59.34%, or 2.978 million tonnes [16]. Production has grown steadily over successive years, and the sector is a significant contributor to agricultural gross domestic product and to employment [16]. Global production trends are consistent with continued expansion of aquaculture as the dominant source of aquatic animals for human consumption [2]. Scales are removed at landing centres, wholesale and retail markets and processing plants, and are ordinarily discarded with other processing waste. The residue is therefore abundant, continuously generated and currently unpriced.

Two qualifications should be attached, because the observation is frequently stated in a form that overstates its practical significance. First, abundance in aggregate is not the same as availability at a plant gate: the material is generated in small quantities at a very large number of dispersed points, and the cost of assembling it into a specification-grade feedstock is a genuine cost rather than a rounding error. Second, mixed-species landings produce a mixed-species scale stream, which propagates the compositional variability discussed in Section 2.1 directly into the process feed. Bangladeshi aquaculture is concentrated in a small number of dominant species, including rohu (Labeo rohita), tilapia and pangas [16], which is favourable for segregation, but the quantity of scale material actually recoverable has not, to the authors’ knowledge, been estimated in the published literature.

Proposed by the authors. Because the residue arises at many dispersed points, a decentralised model — in which washing, drying and size reduction are performed near the point of collection, with stabilised material consolidated at regional processing hubs — is likely to be more workable than centralised collection of wet material. Preliminary drying reduces transport mass and microbial degradation, both of which otherwise erode feedstock quality. This proposal has not been evaluated against alternatives and no collection-cost analysis is offered.

An unresolved issue deserves emphasis. Feedstock quality for biomedical-grade output depends on the time and temperature between removal of the scale and its stabilisation, and on the segregation of species. Neither is currently managed in Bangladeshi fish handling, and neither is addressed in the reviewed literature.

9.2. Domestic Research Base and Processing Capability

Supported by data. Work on scale-derived materials from locally landed species has begun: chitosan has been extracted and characterised from Labeo rohita scales in the South Asian context, with infrared, diffraction, thermal and morphological characterisation reported [14]. Comparable work on locally available substrates elsewhere demonstrates that the analytical methods required are within the reach of university laboratories [13,15]. Published work on fish-scale biomaterials generally, however, is overwhelmingly at laboratory scale, and studies addressing pilot operation, continuous processing or closed mass balance are scarce. Thermal routes to hydroxyapatite use equipment types already present in general manufacturing, which lowers the barrier relative to biochemical extraction [4].

Proposed by the authors. A phased pathway is appropriate, in which locally landed species are first characterised compositionally, a single product line is validated at pilot scale, and additional fractions are added only after the primary line holds specification. The sequence and the decision criterion applicable at each stage are set out in Figure 9b. This is a proposal informed by the reviewed literature, not an evaluated implementation plan, and the criteria are qualitative.

9.3. Market Positioning

Supported by data. Demand exists for marine collagen and hydroxyapatite across biomedical, cosmetic and nutraceutical applications, and fish-derived collagen is favoured in segments where mammalian sources are unacceptable on zoonotic-risk, cultural or religious grounds [1,3,8].

Proposed by the authors. Entry is more plausible through characterised intermediate materials — collagen powders, hydrolysates and hydroxyapatite powders sold to formulators — than through finished medical devices, because the regulatory burden on intermediates is lower and the required quality infrastructure is correspondingly smaller. Movement toward finished products would follow the establishment of quality systems rather than precede it.

No market size, price or share estimate is offered in this review. The reviewed sources do not contain figures that would support such an estimate, and a quantitative market claim should be traceable to a citable market source rather than asserted.

9.4. Enabling Conditions

Materials intended for biomedical, pharmaceutical, cosmetic or food-related use must meet defined specifications for identity, purity, contaminant limits and biological safety, verified by validated analytical methods. For fish-scale-derived products the parameters of interest include, for collagen, molecular weight distribution, helix integrity, and endotoxin and heavy-metal limits; and for hydroxyapatite, phase purity, crystallinity, particle size distribution and trace element content.

Investment in accredited testing capability and in trained analytical personnel is a prerequisite rather than a later refinement, because without it neither process optimisation nor regulatory submission is possible. In the authors’ assessment this is a more binding constraint than access to extraction technology. The enabling conditions required in parallel with technical development are summarised in Figure 9b.

Diverting scales from disposal to processing would create activity along the collection and preprocessing chain and would reduce the organic load presently entering waste streams at markets and processing sites. These are plausible benefits, but their magnitude depends on the volume actually diverted, which is unknown, and on the effluent generated in processing, which is not negligible. Acid demineralisation, alkaline treatment and wash steps generate effluents with substantial chemical and organic loads, and thermal routes carry an energy burden. Where these are untreated, valorisation can shift an environmental burden rather than remove it. The claim that fish-scale valorisation is environmentally beneficial should therefore be made only where effluent treatment is included in the process design and, ideally, where it is supported by life-cycle assessment.

10. Regulatory and Translational Considerations

Three regulatory issues bear specifically on fish-scale-derived materials and are not adequately treated in the applied literature.

The first is classification. A decellularised fish-scale corneal matrix is a xenogeneic tissue-derived implant, and the applicable framework depends on jurisdiction and on whether the material is regarded as a device incorporating a tissue derivative or as a tissue-engineered product. The classification determines the evidence required, and it is not resolved by demonstrating biocompatibility alone. Biomaterial intermediates sold to formulators occupy a different and considerably lighter position, which is the basis for the market-entry proposal in Section 9.3.

The second is source control. Because the feedstock is an aquatic organism, contaminant accumulation in the source animal is a foreseeable hazard, and the dossier for an implantable product would be expected to address heavy metals, residual veterinary compounds and microbiological burden, with traceability to farm or landing site. Farmed material under controlled conditions is more defensible in this respect than material of mixed and unrecorded origin, which bears directly on the collection model discussed in Section 9.1.

The third is process residues and sterilisation. Decellularisation reagents and crosslinkers require toxicological justification, and the sterilisation step must be validated on the finished construct rather than assumed, for the thermal-stability reasons set out in Section 3.3. Characterisation performed before sterilisation does not describe the product that would be implanted.

11. Research Gaps and Future Directions

The obstacles to adoption are, in the authors’ assessment, better characterised as gaps in process and quality data than as deficiencies in materials science. They are summarised in Table 8, and the principal priorities follow in approximate order of dependency.

Species-resolved compositional characterisation. Collagen-to-mineral ratio, lamellar organisation and denaturation temperature vary with species, season and habitat [6,9], yet systematic comparative data across commercially significant species are lacking. Without them, neither species segregation nor process design tolerant of a defined input range is possible.

A minimum reporting standard. Comparability across published extraction studies is poor: yields are reported on inconsistent bases, purity criteria differ, and structural integrity is not always verified. Adoption of a common minimum reporting set — feedstock species and provenance, basis of yield calculation, verification of helix integrity, phase identification for mineral products, and contaminant screening — would materially improve the usefulness of subsequent work. In the authors’ assessment this is the single change that would most improve the field.

Pilot-scale demonstration with closed mass balance. The literature terminates at laboratory scale. Continuous or semi-continuous operation with a closed mass balance, reagent and energy inventories, and effluent characterisation is the missing evidence on which any techno-economic or life-cycle assessment depends.

Techno-economic and life-cycle assessment. Both are largely absent. Assessments should include collection and stabilisation costs, reagent and energy consumption, effluent treatment, quality control, and the counterfactual of current disposal.

Structural verification of the chitosan pathway. As set out in Section 2.2, the origin and mass balance of the reported chitinous fraction require confirmation by methods that discriminate chitin polymorphs from deacetylated protein residues.

Validation of terahertz characterisation against reference methods. For terahertz measurement to function as a quality-control tool rather than a descriptive one, reproducibility across specimens and instruments must be demonstrated, a reference spectrum for a material of known specification established, and the sensitivity of the response to process variables determined, with calibration against the methods in Table 4.

Progression of the ocular evidence base. The preclinical record for scale-derived corneal matrices is comparatively strong but stops short of clinical evaluation [31,32,33,34,35]. The intermediate steps are a defined release specification, validated sterilisation, and longer-duration large-animal studies with optical as well as tectonic endpoints.

12. Conclusions

Fish scales are a defensible feedstock for regenerative biomaterials, but the case for them is narrower and more conditional than it is usually presented. Two recovery routes are well supported — Type I collagen from the organic matrix and hydroxyapatite from the mineral phase — both reported across multiple species and laboratories. A third, chitosan from a chitinous scale fraction, has an empirical literature but lacks confirmation of the fraction’s origin and mass balance, and should not be planned upon as though it were equivalent.

The distinguishing feature of scales relative to other fisheries residues is the co-presence of organic and mineral fractions, which permits cascading recovery. That advantage is real but partial: the separation chemistries compete, so each fraction is recovered at less than its standalone optimum, and every added recovery step increases capital, energy and effluent burden. Integration is a design decision to be justified case by case, not a general principle.

In ophthalmology, the scale offers something more specific than collagen content. Its base plate is already organised into oriented lamellae, which is the property a corneal stromal substitute requires and which reconstituted collagen must be engineered to acquire. Decellularised scale-derived matrices have accumulated a coherent preclinical record spanning in vitro cytocompatibility with human corneal cells and implantation in rats, rabbits and minipigs, which places them ahead of most candidate corneal biomaterials. They have not entered controlled clinical evaluation, and the barriers to doing so are geometric constraint, a narrow thermal processing window, residual-reagent justification and the absence of a release specification — not any deficiency in the underlying biology.

The regenerative capacity of the teleost retina is a separate matter and should be kept separate. It is a model of endogenous repair through Müller glial reprogramming, pursued in host tissue by pharmacological and genetic means. It involves no fish-derived implant, and fish-scale collagen has no demonstrated capacity to induce retinal neurogenesis. Both strands take their cue from teleost biology and both address vision loss, but they converge only at the level of clinical intent.

Terahertz methods offer a label-free, non-destructive and hydration-sensitive probe that is well suited to comparative screening and sub-surface imaging, and plausibly to inline monitoring of drying and demineralisation. On present evidence they do not substitute for infrared spectroscopy, diffraction, calorimetry or electrophoresis in establishing the properties that determine whether a collagen preparation is fit for use, and absorbance data do not report mechanical performance. Their realistic role is complementary and calibrated.

For producer economies, feedstock is not the constraint. The constraints are the logistics of assembling dispersed residue into a specification-grade feed, the absence of pilot-scale process data, limited accredited analytical capacity, undefined regulatory pathways for biomaterial intermediates, and the absence of techno-economic and life-cycle evidence. These are engineering, analytical and institutional problems rather than materials-chemistry problems, and progress in extraction methods alone will not resolve them. A realistic route begins with characterised intermediate materials, supported by species-resolved compositional data and pilot validation, with movement into biomedical grades following the establishment of quality systems rather than preceding it. On that basis, fish-scale valorisation offers a credible but demanding opportunity within the circular bioeconomy.