Zahra Taghados - Department of Computer Science,Engineering and InformationTechnologyShiraz UniversityShiraz, Iran
Ramin Takmil - Department of Computer Science,Engineering and InformationTechnologyShiraz UniversityShiraz, Iran
Zohreh Azimifar - Department of Computer Science,Engineering and InformationTechnologyShiraz UniversityShiraz, Iran
Malihezaman Monsefi - Department of BiologyShiraz UniversityShiraz, Iran

Medical image classification is a critical task inmodern healthcare, aiding in accurate diagnoses and treatmentplanning. With the advent of Machine Learning (ML) and DeepLearning (DL) techniques, significant progress has been madein automating this process. However, traditional ML and DLmodels often lack the capacity to infer causal relationships,which limits their interpretability and generalizability. Thispaper presents a comprehensive analysis of the integration ofcausality in medical image classification, particularyly,integration of causality in deep learning for the classification,utilizing two distinct datasets: cervical cell images and CT scansof leg bones. Through rigorous evaluation, our approachdemonstrates promising results with an accuracy of ۹۷.۵% forcervical cell image classification and ۹۲.۵% for bone CT scans.This study underscores the transformative potential ofcausality-driven techniques in advancing reliable clinicaldecision support systems.

Medical Image Classification, MachineLearning, Deep Learning, Causality, Causal Inference