Talukdar, R. | bioRxiv Science

bioRxiv2024

Majority-LCL: Towards Malaria Cell Detection using Label Contrastive Learning and Majority voting Ensembling

Malaria is a contagious disease caused by Plasmodium, a group of single-celled parasites, and is most commonly transmitted by an infected female Anopheles mosquito. More than 40 percent of the global population is at risk, with approximately 219 million reported cases and around 435000 deaths recorded in 2017 alone. Despite the availability of several advanced diagnostic tools, accurate malaria diagnosis remains challenging in resource-constrained settings, where microscopists often struggle to improve diagnostic accuracy. Deep learning-based cell image classification helps reduce incorrect diagnostic conclusions by enabling automated analysis. This research aims to improve diagnostic accuracy by classifying malaria-infected cells using a majority voting ensemble framework combined with triplet loss aided label contrastive learning. Experimental results demonstrate the effectiveness of the proposed method on microscopic cell images in terms of accuracy, precision, recall, and other evaluation metrics.

bioinformatics↗

bioRxiv2024

RAF2Net: Automated grading of Renal cell Carcinoma utilizing Attention-enhanced deep learning models through Feature Fusion

It is anticipated that the number of instances of kidney cancer will continue to rise globally, which motivates changes to the current diagnostic framework in order to address emerging issues. Renal cell carcinoma (RCC) accounts for 80-85% of all renal tumors and is the most common kind of kidney cancer. Based on kidney histopathology images, this study presented a completely automated, robust, and computationally efficient Renal Cell Carcinoma Grading Network (RAF 2Net). Our suggested model incorporates 3 different Mobilenet backbones with intelligent feature fusion. Moreover, the attention blocks help us give more importance to the important pixels, which are majorly responsible for classification. For comparison purposes, Similar tests were conducted using transfer learning methods with pre-trained ImageNet weights and deep learning models created from scratch. To show the efficacy of the suggested method, we have computed evaluation parameters like Accuracy, Precision, F score, Recall, Confusion Matrix, and TSNE. Based on the provided KMC dataset, the experimental result demonstrates that the proposed RAF 2Net outperforms the nine most recent classification methods regarding prediction Accuracy, Recall, Precision, and F score with a value greater than 92%.

bioinformatics↗