Technical considerations of multi-parametric tissue outcome prediction methods in acute ischemic stroke patients

Abstract

Decisions regarding acute stroke treatment rely heavily on imaging, but interpretation can be difficult for physicians. Machine learning methods can assist clinicians by providing tissue outcome predictions for different treatment approaches based on acute multi-parametric imaging. To produce such clinically viable machine learning models, factors such as classifier choice, data normalization, and data balancing must be considered. This study gives comprehensive consideration to these factors by comparing the agreement of voxel-based tissue outcome predictions using acute imaging and clinical parameters with manual lesion segmentations derived from follow-up imaging. This study considers random decision forest, generalized linear model, and k-nearest-neighbor machine learning classifiers in conjunction with three data normalization approaches (non-normalized, relative to contralateral hemisphere, and relative to contralateral VOI), and two data balancing strategies (full dataset and stratified subsampling). These classifier settings were evaluated based on 90 MRI datasets from acute ischemic stroke patients. Distinction was made between patients recanalized using intraarterial and intravenous methods, as well as those without successful recanalization. For primary quantitative comparison, the Dice metric was computed for each voxel-based tissue outcome prediction and its corresponding follow-up lesion segmentation. It was found that the random forest classifier outperformed the generalized linear model and the k-nearest-neighbor classifier, that normalization did not improve the Dice score of the lesion outcome predictions, and that the models generated lesion outcome predictions with higher Dice scores when trained with balanced datasets. No significant difference was found between the treatment groups (intraarterial vs intravenous) regarding the Dice score of the tissue outcome predictions.

Machine Learning-Based Prediction of Brain Tissue Infarction in Patients With Acute Ischemic Stroke Treated With Theophylline as an Add-On to Thrombolytic Therapy: A Randomized Clinical Trial Subgroup Analysis

“…This voxelby-voxel analysis of eight imaging parameters for each voxel based on DWI and PWI sequences collected from each patient together with four clinical parameters practically allows to double the sample size by simulating both treatment outcomes for each patient included in this study. The accuracy of the prediction models was acceptable as the mean Dice similarity coefficients comparing the true lesion volumes with the predicted lesion volumes within each group were within the range of results from a recent study applying multi-parametric tissue outcome prediction methods (12).…”

“…The mean Dice similarity coefficient was 0.40 (SD 0.249) for the theophylline prediction model and 0.35 (SD 0.243) for the placebo prediction model, which is in the range of previously described methods (12).…”

“…A stratified, under-sampled training set consisting of all voxels from the follow-up infarct lesion and an equal number of non-lesion voxels randomly sampled from the ipsilateral hemisphere were extracted for each dataset. This approach and feature setup yielded the best results in a previous in-depth technical evaluation (12).…”

“…The patient in whom the algorithm was tested (used for lesion outcome prediction) in each iteration was always excluded from the aggregated training set used for training of the random forest models. Practically, the machine learning models were implemented using ALGLIB (www.alglib.net) (13) with the random forest model consisting of 100 trees assuring an acceptable compromise between accuracy and computation speed, described in more details by Winder et al (12).…”

Treatment Efficacy Analysis in Acute Ischemic Stroke Patients Using In Silico Modeling Based on Machine Learning: A Proof-of-Principle

“…However, no such model exists for the evaluation of stroke treatments due to the high complexity of the cerebrovascular system and interventional procedure, which renders an in silico validation, for example using computational fluid dynamics simulations, infeasible. Instead, we propose the use of machine learning models, which have been used previously to model the evolution of acute ischemic stroke under particular treatment conditions [15][16][17][18][19][20][21][22][23]. In these studies, real patient data were used to empirically and automatically optimize (train) the machine learning model for the purpose of predicting patients' final voxel-wise tissue outcome given their baseline imaging and clinical information.…”