Deep learning improves image quality and radiomics reproducibility for high-speed four-dimensional computed tomography reconstruction

A novel loss function to reproduce texture features for deep learning-based MRI-to-CT synthesis

BACKGROUND

Studies on computed tomography (CT) synthesis based on magnetic resonance imaging (MRI) have mainly focused on pixel-wise consistency, but the texture features of regions of interest (ROIs) have not received appropriate attention.

PURPOSE

This study aimed to propose a novel loss function to reproduce texture features of ROIs and pixel-wise consistency for deep learning-based MRI-to-CT synthesis. The method was expected to assist the multi-modality studies for radiomics.

METHODS

The study retrospectively enrolled 127 patients with nasopharyngeal carcinoma. CT and MRI images were collected for each patient, and then rigidly registered as pre-procession. We proposed a gray-level co-occurrence matrix (GLCM)-based loss function to improve the reproducibility of texture features. This novel loss function could be embedded into the present deep learning-based framework for image synthesis. In this study, a typical image synthesis model was selected as the baseline, which contained a Unet trained mean square error (MSE) loss function. We embedded the proposed loss function and designed experiments to supervise different ROIs to prove its effectiveness. The concordance correlation coefficient (CCC) of the GLCM feature was employed to evaluate the reproducibility of GLCM features, which are typical texture features. Besides, we used a publicly available dataset of brain tumors to verify our loss function.

RESULTS

Compared with the baseline, the proposed method improved the pixel-wise image quality metrics (MAE: 107.5 to 106.8 HU; SSIM: 0.9728 to 0.9730). CCC values of the GLCM features in GTVnx were significantly improved from 0.78 ± 0.12 to 0.82 ± 0.11 (p < 0.05 for paired t-test). Generally, > 90% (22/24) of the GLCM-based features were improved compared with the baseline, where the Informational Measure of Correlation feature was improved the most (CCC: 0.74 to 0.83). For the public dataset, the loss function also shows its effectiveness. With our proposed loss function added, the ability to reproduce texture features was improved in the ROIs.

CONCLUSIONS

The proposed method reproduced texture features for MRI-to-CT synthesis, which would benefit radiomics studies based on image multi-modality synthesis.

Reproducibility in Radiomics: A Comparison of Feature Extraction Methods and Two Independent Datasets

Radiomics involves the extraction of information from medical images that are not visible to the human eye. There is evidence that these features can be used for treatment stratification and outcome prediction. However, there is much discussion about the reproducibility of results between different studies. This paper studies the reproducibility of CT texture features used in radiomics, comparing two feature extraction implementations, namely the MATLAB toolkit and Pyradiomics, when applied to independent datasets of CT scans of patients: (i) the open access RIDER dataset containing a set of repeat CT scans taken 15 min apart for 31 patients (RIDER Scan 1 and Scan 2, respectively) treated for lung cancer; and (ii) the open access HN1 dataset containing 137 patients treated for head and neck cancer. Gross tumor volume (GTV), manually outlined by an experienced observer available on both datasets, was used. The 43 common radiomics features available in MATLAB and Pyradiomics were calculated using two intensity-level quantization methods with and without an intensity threshold. Cases were ranked for each feature for all combinations of quantization parameters, and the Spearman's rank coefficient, rs, calculated. Reproducibility was defined when a highly correlated feature in the RIDER dataset also correlated highly in the HN1 dataset, and vice versa. A total of 29 out of the 43 reported stable features were found to be highly reproducible between MATLAB and Pyradiomics implementations, having a consistently high correlation in rank ordering for RIDER Scan 1 and RIDER Scan 2 (rs > 0.8). 18/43 reported features were common in the RIDER and HN1 datasets, suggesting they may be agnostic to disease site. Useful radiomics features should be selected based on reproducibility. This study identified a set of features that meet this requirement and validated the methodology for evaluating reproducibility between datasets.

In Vivo Repeatability and Multiscanner Reproducibility of MRI Radiomics Features in Patients With Monoclonal Plasma Cell Disorders: A Prospective Bi-institutional Study

OBJECTIVES

Despite the extensive number of publications in the field of radiomics, radiomics algorithms barely enter large-scale clinical application. Supposedly, the low external generalizability of radiomics models is one of the main reasons, which hinders the translation from research to clinical application. The objectives of this study were to investigate reproducibility of radiomics features (RFs) in vivo under variation of patient positioning, magnetic resonance imaging (MRI) sequence, and MRI scanners, and to identify a subgroup of RFs that shows acceptable reproducibility across all different acquisition scenarios.

MATERIALS AND METHODS

Between November 30, 2020 and February 16, 2021, 55 patients with monoclonal plasma cell disorders were included in this prospective, bi-institutional, single-vendor study. Participants underwent one reference scan at a 1.5 T MRI scanner and several retest scans: once after simple repositioning, once with a second MRI protocol, once at another 1.5 T scanner, and once at a 3 T scanner. Radiomics feature from the bone marrow of the left hip bone were extracted, both from original scans and after different image normalizations. Intraclass correlation coefficient (ICC) was used to assess RF repeatability and reproducibility.

RESULTS

Fifty-five participants (mean age, 59 ± 7 years; 36 men) were enrolled. For T1-weighted images after muscle normalization, in the simple test-retest experiment, 110 (37%) of 295 RFs showed an ICC ≥0.8: 54 (61%) of 89 first-order features (FOFs), 35 (95%) of 37 volume and shape features, and 21 (12%) of 169 texture features (TFs). When the retest was performed with different technical settings, even after muscle normalization, the number of FOF/TF with an ICC ≥0.8 declined to 58/13 for the second protocol, 29/7 for the second 1.5 T scanner, and 49/7 for the 3 T scanner, respectively. Twenty-five (28%) of the 89 FOFs and 6 (4%) of the 169 TFs from muscle-normalized T1-weighted images showed an ICC ≥0.8 throughout all repeatability and reproducibility experiments.

CONCLUSIONS

In vivo, only few RFs are reproducible with different MRI sequences or different MRI scanners, even after application of a simple image normalization. Radiomics features selected by a repeatability experiment only are not necessarily suited to build radiomics models for multicenter clinical application. This study isolated a subset of RFs, which are robust to variations in MRI acquisition observed in scanners from 1 vendor, and therefore are candidates to build reproducible radiomics models for monoclonal plasma cell disorders for multicentric applications, at least when centers are equipped with scanners from this vendor.