Sourav Basak Shuvo

Medical image classifiers detect gastrointestinal diseases well, but they do not explain their decisions. Large language models can generate clinical text, yet they struggle with visual reasoning and often produce unstable or incorrect explanations. This leaves a gap between what a model sees and the type of reasoning a clinician expects. We introduce a framework that links image classification with structured clinical reasoning. A new hybrid model, MobileCoAtNet, is designed for endoscopic images and achieves high accuracy across eight stomach-related classes. Its outputs are then used to drive reasoning by several LLMs. To judge this reasoning, we build two expert-verified benchmarks covering causes, symptoms, treatment, lifestyle, and follow-up care. Thirty-two LLMs are evaluated against these gold standards. Strong classification improves the quality of their explanations, but none of the models reach human-level stability. Even the best LLMs change their reasoning when prompts vary. Our study shows that combining DL with LLMs can produce useful clinical narratives, but current LLMs remain unreliable for high-stakes medical decisions. The framework provides a clearer view of their limits and a path for building safer reasoning systems. The complete source code and datasets used in this study are available at https://github.com/souravbasakshuvo/DL3M.

This paper presents a hybrid deep learning model, TransfusionNet, comprising VGG19, ResNet50, and ViT with fusion for early detection of cervical cancer. An improved medical image preprocessing technique combining contrast-limited adaptive histogram equalization (CLAHE), Laplacian sharpening, bilateral filtering, and unsharp masking has been applied to enhance image quality. An early-layer feature fusion strategy was used to enhance feature diversity and improve decision-making, unlike conventional methods that fuse only at final layers. The fusion strategy combines features from VGG19, ResNet50, and ViT at multiple stages using concatenation and element-wise addition, allowing integration of detailed and high-level information across different levels. Two separate transition blocks were used to align VGG19 and ResNet50 outputs at different layers. TransfusionNet is trained and tested on the publicly available SIPaKMeD dataset, containing 4,049 images of five classes. TransfusionNet is also verified on Herlev and LCPSI datasets. The five-fold cross-validation was employed to mitigate overfitting and enhance model’s robustness and generalizability. TransfusionNet demonstrates promising results for early cervical cancer detection, achieving state-of-the-art performance on the SIPaKMeD dataset. For batch size 128, TransfusionNet achieved an average accuracy of 99.40%, an F1-score of 99.36%, sensitivity of 99.25%, and Cohen’s Kappa of 99.67% across the five folds.

Bioinformatics interprets biological data, especially genomics, to reveal shared genetic and molecular traits across cancers. Background research demonstrates that lung, cervical, liver, and prostate cancers share a significant number of biochemical and genetic traits. To conduct this study, we collected gene expression datasets: GSE19188 (lung cancer), GSE14520 (liver cancer), GSE46602 (prostate cancer), and GSE63514 (cervical cancer) from the Gene Expression Omnibus (GEO) database hosted by NCBI. The GEO2R tool was used for initial preprocessing, filtering, and identification of differentially expressed genes (DEGs). A total of 39 common DEGs shared across all four cancers were identified, from which 10 hub genes (PIK3R1, EZH2, ECT2, KPNA2, CCNB1, AURKA, CLU, TOP2A, BUB1B, ETS2) were selected based on degree centrality within the protein-protein interaction (PPI) network. To understand the biological relevance of these genes, gene ontology (GO), pathway enrichment, and heatmap analysis were performed. Additionally, regulatory and interaction networks were constructed, including the PPI network, TF-gene regulatory network, gene-miRNA interaction network, and protein-chemical interaction network, based on the hub genes. This network-based analysis revealed 15 key miRNAs and 17 transcription factors (TFs) as major regulatory components. Finally, potential therapeutic drug candidates were predicted through drug enrichment analysis using Enrichr, by examining their interactions with the key hub genes. In future, we aim to apply molecular docking and simulation techniques using plant-derived compounds to develop a unified therapeutic strategy targeting all four cancers.

Gallbladder cancer is a highly challenging malignancy within the field of oncology, distinguished by its diagnosis at an advanced stage and its restricted rates of survival. Gallbladder cancer classification from ultrasound images is a very difficult task due to poor image quality and speckle artifact. We suggest employing an ensemble of well-known Convolutional Neural Network models, such as VGG16, VGG19, XceptionNet, and ResNet50, to evaluate their effectiveness in differentiating between normal, benign, and malignant gall bladder conditions. As a result, the ensemble deep learning model significantly enhances the accuracy of classification.