{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,15]],"date-time":"2026-07-15T05:05:40Z","timestamp":1784091940938,"version":"3.55.0"},"reference-count":35,"publisher":"MDPI AG","issue":"14","license":[{"start":{"date-parts":[[2026,7,13]],"date-time":"2026-07-13T00:00:00Z","timestamp":1783900800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Diagnostics"],"abstract":"<jats:p>Background\/Objectives: Colorectal cancer (CRC) is the second leading cause of cancer-related mortality worldwide, with polyp miss rates of up to 26% reported during colonoscopy and classification accuracy remaining highly operator-dependent. Accurate multi-class polyp subtype classification is clinically critical, as it directly determines treatment decisions: adenomatous polyps require resection, whereas hyperplastic lesions may warrant only surveillance. This study aims to systematically compare five deep learning architectures for five-class gastrointestinal polyp classification and to provide clinically interpretable diagnostic insights through Grad-CAM visualisation. Methods: ResNet50, VGG16, EfficientNet-B3, DenseNet121, and Vision Transformer (ViT-B\/16) were evaluated on the Kvasir Dataset V2 (5000 images, five classes) under a unified training and evaluation protocol on common GPU hardware. All models employed ImageNet transfer learning with a redesigned multi-layer classification head. Two optimisation strategies were applied: SGD with cosine annealing for CNN architectures, and AdamW with linear warmup for ViT-B\/16. Gradient-weighted Class Activation Mapping (Grad-CAM) was applied to generate spatial attention heatmaps for qualitative clinical interpretation. Results: Under a single 80\/10\/10 split, ViT-B\/16 attained the highest accuracy (97.2%); however, because a single split is sensitive to sampling, the evaluation was strengthened with stratified five-fold cross-validation (mean \u00b1 SD). Under cross-validation, EfficientNet-B3 achieved the highest accuracy at 95.90 \u00b1 0.35%, followed closely by ViT-B\/16 (95.12 \u00b1 0.72%), then ResNet50 (91.74 \u00b1 0.74%), DenseNet121 (90.32 \u00b1 0.70%), and VGG16 (88.50 \u00b1 1.72%); the small standard deviations indicate that all models, including ViT-B\/16, were stable across folds. Pairwise McNemar tests with Holm correction found that every difference was statistically significant (p &lt; 0.05), including the EfficientNet-B3 advantage over ViT-B\/16 (p = 0.010). ViT-B\/16 thus remained a strong, stable performer that significantly outperformed the three remaining CNNs, while the cross-validated ranking placed the most compact model, EfficientNet-B3, first: a Vision Transformer was highly competitive with, but not superior to, the strongest CNN. A consistent, architecture-agnostic misclassification pattern was identified between dyed-lifted polyps and dyed-resection margins across all five models, consistent with a task-level visual ambiguity that may also reflect overlapping class definitions and annotation factors, with direct clinical implications. Grad-CAM analysis, quantified by attention-entropy and concentration metrics, showed that model attention remained focused on relevant stained tissue regardless of whether predictions were correct, indicating that the dyed-class confusions reflect genuine visual ambiguity rather than a localisation failure. Conclusions: Under cross-validation, EfficientNet-B3 achieved the highest accuracy on the Kvasir V2 five-class task, significantly outperforming all other architectures, with ViT-B\/16 being a close and competitive second. The identified confusion between post-procedural chromoendoscopic classes is unlikely to be fully resolved by architectural changes alone and may require higher-resolution imaging or domain expert re-annotation. These findings contribute to the evidence base for explainable deep learning in gastrointestinal endoscopy; external, multi-centre validation remains necessary before clinical adoption.<\/jats:p>","DOI":"10.3390\/diagnostics16142182","type":"journal-article","created":{"date-parts":[[2026,7,14]],"date-time":"2026-07-14T00:55:21Z","timestamp":1783990521000},"page":"2182","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Optimised Deep Learning for Gastrointestinal Polyp Classification: A Controlled Benchmark of Five CNN and Transformer Architectures with Grad-CAM Interpretability"],"prefix":"10.3390","volume":"16","author":[{"given":"Zhengsui","family":"Gu","sequence":"first","affiliation":[{"name":"School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China"},{"name":"Department of Engineering, College of Engineering, Brunel University of London, Uxbridge UB8 3PH, UK"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hoda Anwar","family":"Ibrahim","sequence":"additional","affiliation":[{"name":"Department of Engineering, College of Engineering, Brunel University of London, Uxbridge UB8 3PH, UK"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wamadeva","family":"Balachandran","sequence":"additional","affiliation":[{"name":"Department of Engineering, College of Engineering, Brunel University of London, Uxbridge UB8 3PH, UK"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Md Nazmul","family":"Huda","sequence":"additional","affiliation":[{"name":"Department of Engineering, College of Engineering, Brunel University of London, Uxbridge UB8 3PH, UK"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,7,13]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","unstructured":"Maida, M., Dahiya, D.S., Shah, Y.R., Tiwari, A., Gopakumar, H., Vohra, I., Khan, A., Jaber, F., Ramai, D., and Facciorusso, A. (2024). Screening and surveillance of colorectal cancer: A review of the literature. Cancers, 16.","DOI":"10.3390\/cancers16152746"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"844","DOI":"10.1053\/j.gastro.2012.06.001","article-title":"Guidelines for Colonoscopy Surveillance After Screening and Polypectomy: A Consensus Update by the US Multi-Society Task Force on Colorectal Cancer","volume":"143","author":"Lieberman","year":"2012","journal-title":"Gastroenterology"},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Pogorelov, K., Randel, K.R., Griwodz, C., and de Lange, T. (2017, January 20\u201323). KVASIR: A multi-class image dataset for computer aided gastrointestinal disease detection. Proceedings of the 8th ACM Multimedia Systems Conference (MMSys), Taipei, Taiwan.","DOI":"10.1145\/3083187.3083212"},{"key":"ref_4","first-page":"1289","article-title":"Review of application of U-Net and Transformer in colon polyp image segmentation","volume":"42","author":"Shi","year":"2025","journal-title":"Sheng Wu Yi Xue Gong Cheng Xue Za Zhi"},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"Krenzer, A., Heil, S., Fitting, D., Matti, S., Zoller, W.G., Hann, A., and Puppe, F. (2023). Automated classification of polyps using deep learning architectures and few-shot learning. BMC Med. Imaging, 23.","DOI":"10.1186\/s12880-023-01007-4"},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"396","DOI":"10.1007\/s10462-025-11369-6","article-title":"Advancement in semantic segmentation techniques: A comprehensive review for semantic segmentation of colorectal polyps using deep learning","volume":"58","author":"Manan","year":"2025","journal-title":"Artif. Intell. Rev."},{"key":"ref_7","doi-asserted-by":"crossref","unstructured":"Lin, L., Huang, W., and Ouyang, N. (2025). GFANet: Global feature attention network for polyp segmentation. J. Digit. Imaging Inform. Med.","DOI":"10.1007\/s10278-025-01734-w"},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"45484","DOI":"10.1038\/s41598-025-29566-2","article-title":"Vision language models versus machine learning models performance on polyp detection and classification in colonoscopy images","volume":"15","author":"Khalafi","year":"2025","journal-title":"Sci. Rep."},{"key":"ref_9","first-page":"658","article-title":"Exploring the effectiveness of deep learning architectures for colorectal polyp detection: Performance analysis and insights","volume":"5","author":"Neamah","year":"2024","journal-title":"SN Comput. Sci."},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"106451","DOI":"10.1016\/j.bspc.2024.106451","article-title":"Comparative analysis of machine learning frameworks for automatic polyp characterization","volume":"95","author":"Jain","year":"2024","journal-title":"Biomed. Signal Process. Control"},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Kim, H.E., Cosa-Linan, A., Santhanam, N., Jannesari, M., Maros, M.E., and Ganslandt, T. (2022). Transfer learning for medical image classification: A literature review. BMC Med. Imaging, 22.","DOI":"10.1186\/s12880-022-00793-7"},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Wang, W., He, J., and Wang, X. (2024, January 5\u20138). DFE-IANet: A method for polyp image classification based on dual-domain feature extraction and interaction attention. Proceedings of the International Conference on Intelligent Computing (ICIC), Tianjin, China.","DOI":"10.65286\/icic.v20i1.69572"},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"1069","DOI":"10.1053\/j.gastro.2018.06.037","article-title":"Deep learning localizes and identifies polyps in real time with 96% accuracy in screening colonoscopy","volume":"155","author":"Urban","year":"2018","journal-title":"Gastroenterology"},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"568","DOI":"10.1053\/j.gastro.2017.10.010","article-title":"Accurate classification of diminutive colorectal polyps using computer-aided analysis","volume":"154","author":"Chen","year":"2018","journal-title":"Gastroenterology"},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"512","DOI":"10.1053\/j.gastro.2020.04.062","article-title":"Efficacy of real-time computer-aided detection of colorectal neoplasia in a randomized trial","volume":"159","author":"Repici","year":"2020","journal-title":"Gastroenterology"},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"71","DOI":"10.1016\/S2468-1253(18)30282-6","article-title":"Artificial intelligence and computer-aided diagnosis in colonoscopy: Current evidence and future directions","volume":"4","author":"Ahmad","year":"2019","journal-title":"Lancet Gastroenterol. Hepatol."},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"2027","DOI":"10.1053\/j.gastro.2018.04.003","article-title":"Artificial intelligence-assisted polyp detection for colonoscopy: Initial experience","volume":"154","author":"Misawa","year":"2018","journal-title":"Gastroenterology"},{"key":"ref_18","first-page":"1434","article-title":"Cost savings in colonoscopy with artificial intelligence-aided polyp diagnosis: An add-on analysis of a clinical trial","volume":"174","author":"Mori","year":"2021","journal-title":"Ann. Intern. Med."},{"key":"ref_19","first-page":"102350","article-title":"Deep learning for gastrointestinal cancer diagnosis: A systematic survey","volume":"130","author":"Srivastava","year":"2022","journal-title":"Artif. Intell. Med."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"104003","DOI":"10.1016\/j.compbiomed.2020.104003","article-title":"A comprehensive review of deep learning in colon cancer","volume":"126","author":"Pacal","year":"2020","journal-title":"Comput. Biol. Med."},{"key":"ref_21","unstructured":"Houmaidi, W., Hadadi, M., Sabiri, Y., and Chtouki, Y. (2026). DeepGI: Explainable deep learning for gastrointestinal image classification. arXiv."},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Wahid, S.B., Rothy, Z.T., News, R.K., and Rieyan, S.A. (2025). Interpretable deep learning approaches for reliable GI image classification: A study with the HyperKvasir dataset. medRxiv.","DOI":"10.1109\/ICCIT64611.2024.11022387"},{"key":"ref_23","unstructured":"Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., and Gelly, S. (2021, January 3\u20137). An image is worth 16\u00d716 words: Transformers for image recognition at scale. Proceedings of the 9th International Conference on Learning Representations (ICLR), Virtual."},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"103765","DOI":"10.1016\/j.media.2025.103765","article-title":"Collaborative surgical instrument segmentation for monocular depth estimation in minimally invasive surgery","volume":"106","author":"Li","year":"2025","journal-title":"Med. Image Anal."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"9460","DOI":"10.1038\/s41467-024-53826-w","article-title":"Classification of distinct tendinopathy subtypes for precision therapeutics","volume":"15","author":"Tang","year":"2024","journal-title":"Nat. Commun."},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Shen, M.-H., Huang, C.-C., Chen, Y.-T., Tsai, Y.-J., Liou, F.-M., Chang, S.-C., and Phan, N.N. (2023). Deep learning empowers endoscopic detection and polyps classification: A multiple-hospital study. Diagnostics, 13.","DOI":"10.3390\/diagnostics13081473"},{"key":"ref_27","unstructured":"Tan, M., and Le, Q.V. (2019, January 9\u201315). EfficientNet: Rethinking model scaling for convolutional neural networks. Proceedings of the 36th International Conference on Machine Learning (ICML), Long Beach, CA, USA."},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"He, K., Zhang, X., Ren, S., and Sun, J. (2016, January 27\u201330). Deep residual learning for image recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA.","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref_29","unstructured":"Simonyan, K., and Zisserman, A. (2015, January 7\u20139). Very deep convolutional networks for large-scale image recognition. Proceedings of the 3rd International Conference on Learning Representations (ICLR), San Diego, CA, USA."},{"key":"ref_30","doi-asserted-by":"crossref","unstructured":"Huang, G., Liu, Z., van der Maaten, L., and Weinberger, K.Q. (2017, January 21\u201326). Densely connected convolutional networks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, USA.","DOI":"10.1109\/CVPR.2017.243"},{"key":"ref_31","unstructured":"Loshchilov, I., and Hutter, F. (2017, January 24\u201326). SGDR: Stochastic gradient descent with warm restarts. Proceedings of the 5th International Conference on Learning Representations (ICLR), Toulon, France."},{"key":"ref_32","doi-asserted-by":"crossref","unstructured":"Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., and Batra, D. (2017, January 22\u201329). Grad-CAM: Visual explanations from deep networks via gradient-based localization. Proceedings of the IEEE International Conference on Computer Vision (ICCV), Venice, Italy.","DOI":"10.1109\/ICCV.2017.74"},{"key":"ref_33","unstructured":"Ioffe, S., and Szegedy, C. (2015, January 7\u20139). Batch normalization: Accelerating deep network training by reducing internal covariate shift. Proceedings of the 32nd International Conference on Machine Learning (ICML), Lille, France."},{"key":"ref_34","first-page":"14738","article-title":"Brain tumor analysis empowered with deep learning: A review, taxonomy, and prospects","volume":"8","author":"Nadeem","year":"2020","journal-title":"IEEE Access"},{"key":"ref_35","doi-asserted-by":"crossref","first-page":"77","DOI":"10.1016\/j.gie.2020.06.059","article-title":"Performance of artificial intelligence in colonoscopy for adenoma and polyp detection: A systematic review and meta-analysis","volume":"93","author":"Hassan","year":"2021","journal-title":"Gastrointest. Endosc."}],"container-title":["Diagnostics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2075-4418\/16\/14\/2182\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,15]],"date-time":"2026-07-15T04:21:18Z","timestamp":1784089278000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2075-4418\/16\/14\/2182"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,13]]},"references-count":35,"journal-issue":{"issue":"14","published-online":{"date-parts":[[2026,7]]}},"alternative-id":["diagnostics16142182"],"URL":"https:\/\/doi.org\/10.3390\/diagnostics16142182","relation":{},"ISSN":["2075-4418"],"issn-type":[{"value":"2075-4418","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,13]]}}}