eng Oriental Scientific Publishing Company Biomedical and Pharmacology Journal 0974-6242 2026-03-20 19 1 205 232 /10.13005/bpj/3348 70637 article Shagun Sharma 1 Kalpna Guleria 2 Ayush Dogra 3 Monika Sharma 4 Satyam Kumar Agrawal 4 School of Computing Science and Engineering, VIT Bhopal University, Sehore, India Department of Computer Science and Engineering, Chitkara University Institute of Engineering and Technology, Chitkara University, Rajpura, Punjab, India Department of Electronics and Communication Engineering, Chitkara University Institute of Engineering and Technology, Chitkara University, Rajpura, Punjab, India Centre for in Vitro Studies and Translational Research, Chitkara College of Pharmacy, Chitkara University, Rajpura, Punjab, India The manual cell classification methods are highly labour-intensive and time-consuming due to the complex and multidimensional structure of the biological data. The deep learning models are capable of extracting relevant features from the images, processing diverse data types, and handling the nonlinear relationships in the dataset. In the proposed work, two cell lines, MCF7 and HeLa, were cultured in standard conditions and images were captured using a phase contrast microscope. These captured images were grouped on the basis of their confluency and substituted into the deep learning-based Xception, InceptionV3, and MobileNetV2 models for training and multiclass classification. The dataset was augmented with multiple transformations, including rotation, shift, shear, and flipping, which produced 4,375 cancer cell line images. These models were implemented with both the original and augmented datasets, resulting in MobileNetV2 being the most optimal model, producing 94.3% accuracy with the augmented cell lines and 81.5% accuracy with the original cell lines dataset. Further, this model was also found to be the most efficient and lightweight deep learning architecture that provides faster communications in embedded and mobile vision applications. The findings of this work pave the way to further extend the proposed model for the real-time classification and identification of cancer cell lines in in vitro labs. https://biomedpharmajournal.org/vol19no1/label-free-classification-of-mcf7-and-hela-cells-using-high-content-imaging-and-deep-learning/ Cell Classification Deep Learning HeLa High-content Imaging MCF7