Biological tissues may be studied using photoacoustic (PA) spectroscopy, which can yield a wealth of physical and chemical data. However, it is really challenging to directly analyse these tissues because of a lot of data. Data mining techniques can get around this issue. In order to diagnose prostate cancer via PA spectrum assessment, this work describes the machine learning (ML) technique implementation, such as supervised classification and unsupervised hierarchical clustering. The collected PA signals were preprocessed using Pwelch method, and the features are extracted using two methods such as hierarchical cluster and correlation assessment. The extracted features are classified using four ML‐methods, namely, Support Vector Machine (SVM), Naïve Bayes (NB), decision tree C4.5, and Linear Discriminant Analysis (LDA). Furthermore, as these components alter throughout the progression of prostate cancer, this study focuses on the composition and distribution of collagen, lipids, and haemoglobin. In diseased tissues compared to normal tissues, there is a stronger correlation between the various chemical components ultrasonic power spectra, suggesting that the microstructural dispersion in tumour tissues has been more uniform. The accuracy of several classifiers used in cancer tissue diagnosis was greater than 94% for all four methods, which is effective than that of benchmark medical methods. Thus, the method shows significant promise for the noninvasive, early detection of severe prostate cancer.

Breast cancer remains a major cause of mortality among women, where early and accurate detection is critical to improving survival rates. This study presents a hybrid classification approach for mammogram analysis by combining handcrafted statistical features and deep learning techniques. The methodology involves preprocessing with the Shearlet Transform, segmentation using Improved Otsu thresholding and Canny edge detection, followed by feature extraction through Gray Level Co-occurrence Matrix (GLCM), Gray Level Run Length Matrix (GLRLM), and 1st-order statistical descriptors. These features are input into a 2D BiLSTM-CNN model designed to learn spatial and sequential patterns in mammogram images. Evaluated on the MIAS dataset, the proposed method achieved 97.14% accuracy, outperforming several benchmark models. The results indicate that this hybrid strategy offers improvements in classification performance and may assist radiologists in more effective breast cancer screening.

Diabetic Retinopathy (DR) continues to be the leading cause of preventable blindness worldwide, and there is an urgent need for accurate and interpretable framework. A Multi View Cross Attention Vision Transformer (MVCAViT) framework is proposed in this research paper for utilizing the information-complementarity between the dually available macula and optic disc center views of two images from the DRTiD dataset. A novel cross attention-based model is proposed to integrate the multi-view spatial and contextual features to achieve robust fusion of features for comprehensive DR classification. A Vision Transformer and Convolutional neural network hybrid architecture learns global and local features, and a multitask learning approach notes diseases presence, severity grading and lesions localisation in a single pipeline. Results show that the proposed framework achieves high classification accuracy and lesion localization performance, supported by comprehensive evaluations on the DRTiD dataset. Attention-based visualizations further enhance interpretability, indicating the framework’s potential for clinical use. This framework establishes a criterion for improving state-of-the-art retinal image analysis for DR diagnosis which may result in better patient results and final clinical decision.