Submitted:
22 September 2024
Posted:
23 September 2024
You are already at the latest version
Abstract
Fine-needle aspiration cytology (FNAC) is a pivotal diagnostic tool in oncology, utilized for evaluating suspicious lesions and stratifying tumors. The incorporation of machine learning (ML) into FNAC has revolutionized accuracy, efficiency, and diagnostic precision. This comprehensive review explores recent advances in FNAC, emphasizing the transformative role of ML in image-guided sample collection and tumor stratification. Leveraging deep learning and other ML algorithms, researchers have improved diagnostic accuracy, minimized unnecessary biopsies, and optimized treatment selection. This article highlights the transformative applications of machine learning in FNAC while addressing its current limitations and future potential.
Keywords:
Fine Needle Aspiration Cytology (FNAC)
; Machine Learning (ML)
; Deep Learning (DL)
; Image-Guided
; BiopsyTumor Stratification
; Diagnostic Accuracy
; Artificial Intelligence (AI) in Oncology
; Cytopathology
; Predictive Analytics
; Automated Diagnosis
Introduction
Fine-needle aspiration cytology (FNAC) remains an invaluable diagnostic tool for evaluating suspicious lesions and stratifying tumors. Despite being minimally invasive and widely used across various medical specialties, FNAC’s diagnostic precision is sometimes limited due to observer variability, inadequate sampling, and the subjective nature of morphological interpretation. These challenges often lead to unnecessary surgical excisions, repeat procedures, and delays in initiating appropriate therapy.
Machine learning (ML), particularly deep learning (DL), has emerged as a transformative technology in the medical field. Its application in FNAC has significantly improved accuracy, reproducibility, and efficiency. By leveraging ML algorithms for automated and objective analysis of cytological samples, rapid, accurate, and reproducible classification of malignancies is achievable. This review explores how ML is transforming FNAC, particularly in image-guided sample collection and tumor stratification.
Overview of Fine Needle Aspiration Cytology
Technique and Applications
Fine-needle aspiration cytology involves using a thin, hollow needle to obtain cells from a suspicious mass for microscopic examination. The procedure is typically guided by ultrasound (US) or computed tomography (CT) to ensure accurate sampling.
Applications:
- Breast Cancer: FNAC is widely used for evaluating palpable and non-palpable breast lesions.
- Thyroid Cancer: It remains the primary diagnostic tool for thyroid nodules.
- Lymphadenopathy: FNAC plays a significant role in diagnosing lymphoproliferative disorders and metastatic lymph nodes.
- Soft Tissue and Bone Lesions: The procedure is instrumental in evaluating suspicious bone and soft tissue tumors.
- Lung and Mediastinal Lesions: FNAC, often combined with endobronchial ultrasound (EBUS), is used for diagnosing lung and mediastinal lesions.
- Challenges in Traditional FNAC Interpretation:
- Observer Variability: Different pathologists may provide varying interpretations due to subjective judgment.
- Inadequate Sampling: Inadequate cell collection can lead to non-diagnostic results.
- Morphological Overlap: Benign and malignant lesions may share similar cytological features, complicating diagnosis.
Machine Learning in Cytology
Machine learning (ML) is a subset of artificial intelligence that uses algorithms to analyze data patterns and make predictions or classifications. In cytology, ML aids in automated image analysis, enabling precise, objective, and reproducible diagnoses.
Types of Machine Learning:
- Supervised Learning: Algorithms are trained using labeled datasets to classify data or predict outcomes.
- Unsupervised Learning: Identifies hidden patterns or groupings within unlabeled data.
- Reinforcement Learning: Algorithms learn through trial and error to achieve a specific goal.
Deep Learning (DL): A subfield of ML involving neural networks with multiple layers (deep networks). Convolutional Neural Networks (CNNs) are widely used for image classification tasks, making them particularly useful in cytology.
Advances in FNAC Using Machine Learning
Image-Guided Sample Collection
Image-guided FNAC ensures precise and targeted collection of cytological samples. ML algorithms have significantly improved the efficiency and accuracy of this process.
1. Multi-parametric MRI and Glioblastoma
Hu et al. (2015) assessed the feasibility of using machine learning to visualize glioblastoma extent. Their study utilized multi-parametric MRI (mpMRI) and texture analysis to highlight spatial histologic heterogeneity and tumor extent in glioblastoma patients. The authors concluded that machine learning could stratify patients based on the MRI characteristics, thereby aiding personalized treatment decisions (Hu et al., 2015).
2. Image-Guided Raman Spectroscopy for Prostate Cancer
Grajales et al. (2022) proposed an image-guided Raman spectroscopy system for transperineal prostate cancer detection. By combining Raman spectroscopy and multiparametric MRI radiomics, they accurately stratified prostate cancer patients based on their International Society of Urological Pathology (ISUP) grading. This system facilitated precise tumor targeting and improved diagnostic accuracy (Grajales et al., 2022).
3. Machine Learning to Predict Lung Nodule Biopsy Method
Sumathipala et al. (2019) developed a machine learning model using CT image features to predict the most suitable biopsy method for lung nodules, minimizing invasive procedures. Their pilot study showed that machine learning algorithms could distinguish between CT-guided and bronchoscopic biopsy methods, thereby reducing unnecessary procedures (Sumathipala et al., 2019).
4. Rapid Point-of-Care Assessment of Core Needle Cancer Biopsies
Keshavamurthy et al. (2019) described a machine learning-based spectroscopy system for the rapid assessment of core needle biopsies. This system combined spectral data with machine learning algorithms to classify cancer biopsies, providing an accurate, rapid, and non-invasive diagnostic tool (Keshavamurthy et al., 2019).
5. The Cytopathologist in the Hospital-Based FNAC Clinic
Sauer et al. (2018) emphasized the importance of cytopathologists performing their own FNACs using ultrasound (US) guidance. The study outlined how the adoption of US-guided FNAC improved diagnostic accuracy and patient outcomes, enabling cytopathologists to integrate image-guided sampling into their practice (Sauer et al., 2018).
Tumor Stratification Using Machine Learning
Accurate tumor stratification is crucial for personalized therapy. Deep learning and other machine learning algorithms have played a significant role in enhancing stratification accuracy.
1. Artificial Intelligence-Based Assessment of PD-L1 in Lymphoma
Yan et al. (2024) utilized artificial intelligence to assess PD-L1 expression in diffuse large B-cell lymphoma, allowing for precise tumor stratification (Yan et al., 2024).
2. Gleason Scoring for Prostate Core Needle Biopsies
Ao et al. (2023) showed how convolutional neural networks (CNNs) and stimulated Raman scattering microscopy could improve Gleason scoring for prostate biopsies (Ao et al., 2023).
3. Machine Learning in High-Risk Breast Lesions
Bahl et al. (2018) used a machine learning model to predict pathological upgrades and minimize unnecessary surgical excisions of high-risk breast lesions (Bahl et al., 2018).
4. Molecular-Image Cytometry in Oncology
Weissleder and Lee (2020) emphasized the potential of automated molecular-image cytometry in stratifying cancer patients for optimal treatment selection (Weissleder and Lee, 2020).
5. Deep Learning in Breast Biopsy Interpretation
Mullooly et al. (2019) applied deep learning to delineate histologic correlates of mammographic breast density, improving stratification in women referred for diagnostic breast biopsies (Mullooly et al., 2019).
6. Recent Advances and Researches in the Field of Fine Needle Aspiration Cytopathology
Goyal (2023) reviewed recent advances in fine-needle aspiration cytopathology, emphasizing the role of machine learning in improving diagnostic accuracy (Goyal, 2023).
Classification Systems and Automated Diagnosis
1. Sydney System for Lymph Node FNAC
Kanhe et al. (2023) evaluated the Sydney system for stratification of lymph node FNAC. Their retrospective study at a tertiary care center supported the system's accuracy (Kanhe et al., 2023).
2. FNAC Classification of the Breast
Khoury and Sneige (2021) discussed classification approaches for breast FNAC cytopathology, highlighting the advances brought by machine learning (Khoury and Sneige, 2021).
3. Deep Learning in Cytology
Brooks et al. (2020) explored the impact of deep learning on cytology accuracy, providing a framework for future automated diagnostic approaches (Brooks et al., 2020).
4. Automated Molecular-Image Cytometry and Analysis in Modern Oncology
Weissleder and Lee (2020) emphasized the potential of automated molecular-image cytometry in stratifying cancer patients for optimal treatment selection (Weissleder and Lee, 2020).
5. Stimulated Raman Scattering Microscopy Enables Gleason Scoring of Prostate Core Needle Biopsy
Ao et al. (2023) showed how convolutional neural networks (CNNs) and stimulated Raman scattering microscopy could improve Gleason scoring for prostate biopsies (Ao et al., 2023).
6. Recent Advances and Researches in the Field of Fine Needle Aspiration Cytopathology
Goyal (2023) reviewed recent advances in fine-needle aspiration cytopathology, emphasizing the role of machine learning in improving diagnostic accuracy (Goyal, 2023).
Challenges and Future Directions
Despite significant advancements, challenges persist in machine learning for FNAC, including:
- Data Quality and Quantity: Training ML models requires large, high-quality datasets that are often unavailable.
- Generalizability: Models trained on specific datasets may not generalize well to diverse populations.
- Regulatory and Ethical Issues: Ensuring patient privacy and compliance with regulatory standards remains a priority.
Conclusion
Machine learning has revolutionized fine-needle aspiration cytology, improving sample collection and tumor stratification. By leveraging deep learning and other advanced algorithms, researchers have enhanced diagnostic accuracy, minimized unnecessary procedures, and paved the way for personalized cancer treatment. Future research should focus on overcoming current challenges to fully realize the potential of machine learning in FNAC.
References
- Hu LS, Ning S, Eschbacher JM, et al. Multi-parametric MRI and texture analysis to visualize spatial histologic heterogeneity and tumor extent in glioblastoma. PLoS One. 2015;10(11):e0141506. Link.
- Yan F, Da Q, Yi H, et al. Artificial intelligence-based assessment of PD-L1 expression in diffuse large B cell lymphoma. NPJ Precision Oncology. 2024;8(1):5. Link.
- Mullooly M, Ehteshami Bejnordi B, Pfeiffer RM, et al. Application of convolutional neural networks to breast biopsies to delineate tissue correlates of mammographic breast density. NPJ Breast Cancer. 2019;5(1):21. Link.
- Grajales D, Picot F, Shams R, et al. Image-guided Raman spectroscopy navigation system to improve transperineal prostate cancer detection. Part 2: in-vivo tumor-targeting using a convolutional neural network. J Biomed Opt. 2022;27(9):095004. Link.
- Han W, Johnson C, Warner A, et al. Automatic cancer detection on digital histopathology images of mid-gland radical prostatectomy specimens. J Med Imaging. 2020;7(4):047501. Link.
- Kanhe R, Tummidi S, Kothari K, Agnihotri M. Utility of the Proposed Sydney System for Classification of Fine-Needle Aspiration Cytopathology of Lymph Node: A Retrospective Study at a Tertiary Care Center. Acta Cytologica. 2023;67(5):455-464. Link.
- Bahl M, Barzilay R, Yedidia AB, Locascio NJ, Yu L, Lehman CD. High-risk breast lesions: a machine learning model to predict pathologic upgrade and reduce unnecessary surgical excision. Radiology. 2018;286(3):810-818. Link.
- Weissleder R, Lee H. Automated molecular-image cytometry and analysis in modern oncology. Nature Reviews Materials. 2020;5(7):407-421. Link.
- Goldenberg SL, Nir G, Salcudean SE. A new era: artificial intelligence and machine learning in prostate cancer. Nat Rev Urol. 2019;16(7):391-403. Link.
- Keshavamurthy KN, Dylov DV, Yazdanfar S, Patel D. Spectroscopy and machine learning-based rapid point-of-care assessment of core needle cancer biopsies. bioRxiv. 2019;745158. Link.
- Sumathipala Y, Shafiq M, Bongen E, et al. Machine learning to predict lung nodule biopsy method using CT image features: a pilot study. Curr Probl Diagn Radiol. 2019;48(3):212-219. Link.
- Yeung J, Fotiadis N, Diamantopoulos A, Tutt A. Next generation sequencing and image-guided tissue sampling: a primer for interventional radiologists. J Vasc Interv Radiol. 2023;34(8):1257-1267. Link.
- Sauer T, Doughty RW, Orzsagh V, et al. The cytopathologist in the hospital-based FNAC clinic: US image guidance is our new tool to an even better FNAC practice. Mod J Cytol Histopathol. 2018;2(1):7-13. Link.
- Keshavamurthy KN, Dylov DV, Yazdanfar S. Evaluation of an Integrated Spectroscopy and Classification Platform for Point-of-Care Core Needle Biopsy Assessment: Performance Characteristics from Ex Vivo … J Vasc Interv Radiol. 2022;34(8):1257-1267. Link.
- Ao J, Shao X, Liu Z, et al. Stimulated Raman scattering microscopy enables gleason scoring of prostate core needle biopsy by a convolutional neural network. Cancer Res. 2023;83(4):641-653. Link.
- Roberts R, Siddiqui BA, Subudhi SK, et al. Image-guided biopsy/liquid biopsy. In Image-Guided Diagnosis and Therapy in Prostate Cancer (pp. 381-396). Springer. 2020. Link.
- Pisano ED, Fajardo LL, Caudry DJ, et al. Fine-needle aspiration biopsy of nonpalpable breast lesions in a multicenter clinical trial: results from the radiologic diagnostic oncology group V. Radiology. 2001;219(3):789-797. Link.
- Gupta G, Sharma A, Kamboj M, et al. Role of Pathologist in the Era of Image-Guided and EUS-Guided Aspirations: A 10-Year Study at a Single Tertiary Care Oncology Institute in North India. Acta Cytologica. 2022;66(3):187-196. Link.
- Seviar D, Yousuff M, Chia Z, et al. Image-guided core needle biopsy as the first-line diagnostic approach in lymphoproliferative disorders—A review of the current literature. Eur J Haematol. 2021;107(3):312-322. Link.
- Goyal A. Recent Advances and Researches in the Field of Fine Needle Aspiration Cytopathology. IntechOpen. 2023. Link.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.
