Abstract: Colorectal polyp classification (CPC) is essential for the early diagnosis and prevention of colorectal cancer (CC). However, existing methods often struggle to accurately classify polyps due to variations in size, shape, texture, illuminations and imaging artifacts in colonoscopy images. To address these challenges, this paper proposes COPN-NET, a novel deep learning framework combining Mobile RegNet with a Deep Belief Network (DBN). Adaptive Wiener Star filtering is first applied for noise suppression and contrast enhancement. Multi-scale feature extraction is performed using an optimized Mobile RegNet architecture, followed by classification using DBN to distinguish hyperplastic polyps, adenomas, and normal tissues. Experimental evaluations on public colonoscopy datasets demonstrate an accuracy of 95.8% and sensitivity of 96.1%.

Abstract: Hair and scalp diseases such as alopecia areata, psoriasis, seborrheic dermatitis, and folliculitis affect millions globally and require accurate diagnostic assessment. This research presents SCALP CONVICAP, an intelligent computer-aided diagnostic framework for dermatological imaging. The architecture combines ConvNeXt with Graph Neural Networks and Capsule Networks to capture spatial hierarchical relationships and textural scalp variations. Dermatoscopic scalp images undergo illumination correction and hair artifact removal before deep feature representation. Benchmarked against clinical images, the proposed framework achieves an accuracy of 97.1%, showcasing substantial improvement over standard CNN baselines.

Abstract: Coronary artery disease is a leading cause of cardiovascular mortality worldwide. Three-dimensional reconstruction of coronary vessels from 2D angiographic and computed tomography (CT) images is vital for stenosis quantification and surgical planning. This paper proposes CARDIO-3DNet, an end-to-end 3D deep learning framework for vessel segmentation and surface reconstruction. The model employs a 3D U-Net backbone with residual dual-attention blocks to maintain fine vascular branches and bifurcations while eliminating cardiac motion artifacts. On clinical CT angiography datasets, CARDIO-3DNet obtains a Dice Similarity Coefficient of 0.912 and Hausdorff Distance of 3.4 mm.

Abstract: Idiopathic Pulmonary Fibrosis (IPF) is a chronic, progressive lung disorder characterized by irreversible scarring of lung parenchyma. High-resolution computed tomography (HRCT) is the imaging standard for IPF evaluation, but manual quantification of fibrotic lesions is tedious and subjective. We introduce AUD PFS, an Attention U-Net deep learning framework designed specifically for automated segmentation of pulmonary fibrosis patterns including honeycombing, reticulation, and ground-glass opacities. Attention gates suppress irrelevant background features and highlight diseased lung zones. Evaluated on multi-center HRCT datasets, the model achieves a Dice score of 89.4% and an Intersection-over-Union (IoU) of 81.3%.