Voxel‐wise supervised analysis of tumors with multimodal engineered features to highlight interpretable biological patterns

Abstract

Background: Translation of predictive and prognostic image-based learning models to clinical applications is challenging due in part to their lack of interpretability. Some deep-learning-based methods provide information about the regions driving the model output. Yet, due to the high-level abstraction of deep features, these methods do not completely solve the interpretation challenge. In addition,low sample size cohorts can lead to instabilities and suboptimal convergence for models involving a large number of parameters such as convolutional neural networks. Purpose: Here, we propose a method for designing radiomic models that combines the interpretability of handcrafted radiomics with a sub-regional analysis. Materials and methods: Our approach relies on voxel-wise engineered radiomic features with average global aggregation and logistic regression. The method is illustrated using a small dataset of 51 soft tissue sarcoma (STS) patients where the task is to predict the risk of lung metastasis occurrence during the follow-up period. Results: Using positron emission tomography/computed tomography and two magnetic resonance imaging sequences separately to build two radiomic models, we show that our approach produces quantitative maps that highlight the signal that contributes to the decision within the tumor region of interest. In our STS example, the analysis of these maps identified two biological patterns that are consistent with STS grading systems and knowledge: necrosis development and glucose metabolism of the tumor. Conclusions:We demonstrate how that method makes it possible to spatially and quantitatively interpret radiomic models amenable to sub-regions identification and biological interpretation for patient stratification.

Detection of abnormal extraocular muscles in small datasets of computed tomography images using a three-dimensional variational autoencoder

“…To our knowledge, no previous study has applied unsupervised DL to a small dataset of CT images. Most DL studies involving small numbers of medical images used supervised [22][23][24] , or semi-supervised DL [25][26][27] . These approaches may be used because supervised DL algorithms are expected to identify features that distinguish among data in small datasets; they require definite answers to focus on during the model training.…”

Detection of Abnormal Extraocular Muscles in Small Datasets of Computed Tomography Images Using a Three–dimensional Variational Autoencoder: A Pilot Study

“…To our knowledge, no previous study has applied unsupervised DL to a small dataset of CT images. Most DL studies involving small numbers of medical images used supervised, [19][20][21] or semi-supervised DL. [22][23][24] These approaches may be used because supervised DL algorithms are expected to identify features that distinguish among data in small datasets; they require de nite answers to focus on during the model training.…”

Implementing the Point Spread Function Deconvolution for Better Molecular Characterization of Newly Diagnosed Gliomas: A Dynamic 18F-FDOPA PET Radiomics Study

“…One thousand bootstrap iterations of the distribution of the obtained performance were derived and represented as a mean followed by 95% confidence intervals (CI) of individual metrics. Systematic permutation tests were applied to assess whether the models detected a true class structure in the data and performed significantly better than random guessing [ 28 , 29 ]. One-sided Mann-Whitney U tests were applied on the obtained AUCs for the intergroup comparisons between the different feature sets (independent samples) of each individual PSFd status.…”