{"id":1687,"date":"2026-10-06T19:04:27","date_gmt":"2026-10-06T19:04:27","guid":{"rendered":"https:\/\/www.cs.ubbcluj.ro\/~meco\/a-systematic-analysis-of-data-augmentation-and-training-parameters-for-lightweight-cnn-based-mammographic-lesion-classification-2026\/"},"modified":"2026-10-06T19:30:18","modified_gmt":"2026-10-06T19:30:18","slug":"a-systematic-analysis-of-data-augmentation-and-training-parameters-for-lightweight-cnn-based-mammographic-lesion-classification-2026","status":"publish","type":"post","link":"https:\/\/www.cs.ubbcluj.ro\/~meco\/a-systematic-analysis-of-data-augmentation-and-training-parameters-for-lightweight-cnn-based-mammographic-lesion-classification-2026\/","title":{"rendered":"A Systematic Analysis of Data Augmentation and Training Parameters for Lightweight CNN-Based Mammographic Lesion Classification (2026)"},"content":{"rendered":"<div class=\"entry-content\">\n<p>Hybrid Artificial Intelligence Systems<\/p>\n<h2>Authors<\/h2>\n<p>D. Trandafir, M\u0103d\u0103lina Dicu, Camelia Chira<\/p>\n<h2>Abstract<\/h2>\n<p>Breast cancer remains a major cause of cancer-related mortality among women worldwide, making accurate lesion detection essential for improving survival rates and supporting clinical diagnosis. Although current pretrained models for predicting benign or malignant lesions achieve high accuracy, they often focus on a single lesion type and may not provide the level of visual interpretability required by radiologists. This study investigates an approach that differentiates between calcifications and masses, aiming to improve the specificity of automated diagnostic workflows. In this paper, we conduct a systematic analysis of how factors such as contrast intensity, activation function, filter size, and batch size influence the performance of a custom three-layer convolutional neural network trained on ROI images from the CBIS-DDSM dataset. Additionally, we evaluate several offline and online data augmentation strategies, together with different kernel and dropout configurations. Experimental results show that ROI images are highly sensitive to CLAHE-based contrast transformations, with the model achieving its best performance when trained on unfiltered data. Furthermore, online data augmentation using rotations between and improves model robustness, achieving an accuracy of 86%. A small batch size of 8 also provides stable and consistent results across all evaluation metrics.<\/p>\n<h2>Citation<\/h2>\n<pre class=\"wp-block-preformatted\">@Inproceedings{Trandafir2026ASA,\n author = {D. Trandafir and M\u0103d\u0103lina Dicu and Camelia Chira},\n booktitle = {Hybrid Artificial Intelligence Systems},\n title = {A Systematic Analysis of Data Augmentation and Training Parameters for Lightweight CNN-Based Mammographic Lesion Classification},\n year = {2026}\n}<\/pre>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>Hybrid Artificial Intelligence Systems Authors D. Trandafir, M\u0103d\u0103lina Dicu, Camelia Chira Abstract Breast cancer remains a major cause of cancer-related mortality among women worldwide, making accurate lesion detection essential for improving survival rates and supporting clinical diagnosis. Although current pretrained models for predicting benign or malignant lesions achieve high accuracy, they often focus on a&hellip;<\/p>\n","protected":false},"author":6,"featured_media":1719,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":[],"categories":[4],"tags":[45,11,46],"_links":{"self":[{"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/posts\/1687"}],"collection":[{"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/users\/6"}],"replies":[{"embeddable":true,"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/comments?post=1687"}],"version-history":[{"count":2,"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/posts\/1687\/revisions"}],"predecessor-version":[{"id":1720,"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/posts\/1687\/revisions\/1720"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/media\/1719"}],"wp:attachment":[{"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/media?parent=1687"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/categories?post=1687"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.cs.ubbcluj.ro\/~meco\/wp-json\/wp\/v2\/tags?post=1687"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}