Interpretable Machine Learning with Brain Image and Survival Data

Abstract

Recent developments in research on artificial intelligence (AI) in medicine deal with the analysis of image data such as Magnetic Resonance Imaging (MRI) scans to support the of decision-making of medical personnel. For this purpose, machine learning (ML) algorithms are often used, which do not explain the internal decision-making process at all. Thus, it is often difficult to validate or interpret the results of the applied AI methods. This manuscript aims to overcome this problem by using methods of explainable AI (XAI) to interpret the decision-making of an ML algorithm in the use case of predicting the survival rate of patients with brain tumors based on MRI scans. Therefore, we explore the analysis of brain images together with survival data to predict survival in gliomas with a focus on improving the interpretability of the results. Using the Brain Tumor Segmentation dataset BraTS 2020, we used a well-validated dataset for evaluation and relied on a convolutional neural network structure to improve the explainability of important features by adding Shapley overlays. The trained network models were used to evaluate SHapley Additive exPlanations (SHAP) directly and were not optimized for accuracy. The resulting overfitting of some network structures is therefore seen as a use case of the presented interpretation method. It is shown that the network structure can be validated by experts using visualizations, thus making the decision-making of the method interpretable. Our study highlights the feasibility of combining explainers with 3D voxels and also the fact that the interpretation of prediction results significantly supports the evaluation of results. The implementation in python is available on gitlab as “XAIforBrainImgSurv”.

Research on the Application and Interpretability of Predictive Statistical Data Analysis Methods in Medicine

“…As a result, it is difficult to validate or interpret the results. The paper by Matthias Eder and his colleagues [12] is an interesting practical example of how to overcome this problem by using methods of explainable AI (XAI). The authors present an application of visual explanations to interpret the decision of an ML algorithm in the case of predicting the survival rate of brain tumor patients based on their MRI scans.…”

From Code to Cure: The Impact of Artificial Intelligence in Biomedical Applications

“…Machine-learning and deep-learning models also need a high computing cost during model training. To improve the interpretability and transparency of these black box models, explainable AI models are currently used in various studies [38][39][40][41][42][43][44][45][46][47]. The collaboration of experts from biology, computer science and different fields can improve the transparency of such methods.…”

“…Several explainable AI models have been developed for biomedical applications such as MRI scan images to predict the survival of brain tumours [38], ECG data to predict cardiovascular disorders [39] and risk factor identification of diabetic retinopathy [40]. In these studies, SHAP analysis has been incorporated to the AI models to interpret the outcome of the classifier.…”

Enhancing Brain Tumor Classification with Transfer Learning across Multiple Classes: An In-Depth Analysis

“…Ibrokhiov et al [24], in response to the rising incidence of pneumonia, particularly in the wake of the COVID-19 pandemic, introduced an advanced DL-based computer-aided diagnostic system, leveraging transfer learning and parallel computing techniques with VGG19 and ResNet 50 models, achieving an impressive average classification accuracy of 96.6% on the COVID-QU-Ex dataset. Edgar M et al [25] address the challenge of interpreting machine learning algorithms applied to medical image data, specifically in predicting brain tumor survival rates from MRI scans. By leveraging explainable AI techniques, such as Shapley overlays, in conjunction with CNN and the BraTS 2020 dataset, this research demonstrates the improved interpretability of key features, facilitating expert validation and enhancing the overall evaluation of predictive outcomes.…”