MRI-based radiomics to predict response in locally advanced rectal cancer: comparison of manual and automatic segmentation on external validation in a multicentre study

Breast MRI radiomics and machine learning-based predictions of response to neoadjuvant chemotherapy - How are they affected by variations in tumor delineation?

Manual delineation of volumes of interest (VOIs) by experts is considered the gold-standard method in radiomics analysis. However, it suffers from inter- and intra-operator variability. A quantitative assessment of the impact of variations in these delineations on the performance of the radiomics predictors is required to develop robust radiomics based prediction models. In this study, we developed radiomics models for the prediction of pathological complete response to neoadjuvant chemotherapy in patients with two different breast cancer subtypes based on contrast-enhanced magnetic resonance imaging acquired prior to treatment (baseline MRI scans). Different mathematical operations such as erosion, smoothing, dilation, randomization, and ellipse fitting were applied to the original VOIs delineated by experts to simulate variations of segmentation masks. The effects of such VOI modifications on various steps of the radiomics workflow, including feature extraction, feature selection, and prediction performance, were evaluated. Using manual tumor VOIs and radiomics features extracted from baseline MRI scans, an AUC of up to 0.96 and 0.89 was achieved for human epidermal growth receptor 2 positive and triple-negative breast cancer, respectively. For smoothing and erosion, VOIs yielded the highest number of robust features and the best prediction performance, while ellipse fitting and dilation lead to the lowest robustness and prediction performance for both breast cancer subtypes. At most 28% of the selected features were similar to manual VOIs when different VOI delineation data were used. Differences in VOI delineation affect different steps of radiomics analysis, and their quantification is therefore important for development of standardized radiomics research.

Comparison of conventional MRI analysis versus MRI-based radiomics to predict the circumferential margin resection involvement of rectal cancer

BACKGROUND

To compare the application of conventional MRI analysis and MRI-based radiomics to identify the circumferential resection margin (CRM) status of rectal cancer (RC).

METHODS

A cohort of 301 RC patients with 66 CRM invloved status and 235 CRM non-involved status were enrolled in this retrospective study between September 2017 and August 2022. Conventional MRI characteristics included gender, age, diameter, distance to anus, MRI-based T/N phase, CEA, and CA 19 - 9, then the relevant logistic model (Logistic-cMRI) was built. MRI-based radiomics of rectal cancer and mesorectal fascia were calculated after volume of interest segmentation, and the logistic model of rectal cancer radiomics (Logistic-rcRadio) and mesorectal fascia radiomics (Logistic-mfRadio) were constructed. And the combined nomogram (nomo-cMRI/rcRadio/mfRadio) containing conventional MRI characteristics, radiomics of rectal cancer and mesorectal fascia was developed. The receiver operator characteristic curve (ROC) was delineated and the area under curve (AUC) was calculated the efficiency of models.

RESULTS

The AUC of Logistic-cMRI was 0.864 (95%CI, 0.820 to 0.901). The AUC of Logistic-rcRadio was 0.883 (95%CI, 0.832 to 0.928) in the training set and 0.725 (95%CI, 0.616 to 0.826) in the testing set. The AUCs of Logistic-mfRadio was 0.891 (95%CI, 0.838 to 0.936) in the training set and 0.820 (95%CI, 0.725 to 0.905) in the testing set. The AUCs of nomo-cMRI/rcRadio/mfRadio were the highest in both the training set of 0.942 (95%CI, 0.901 to 0.969) and the testing set of 0.909 (95%CI, 0.830 to 0.959).

CONCLUSION

MRI-based radiomics of rectal cancer and mesorectal fascia showed similar efficacy in predicting the CRM status of RC. The combined nomogram performed better in assessment.

Machine learning in predicting pathological complete response to neoadjuvant chemoradiotherapy in rectal cancer using MRI: a systematic review and meta-analysis

OBJECTIVES

To evaluate the performance of machine learning models in predicting pathological complete response (pCR) to neoadjuvant chemoradiotherapy (nCRT) in patients with rectal cancer using magnetic resonance imaging.

METHODS

We searched PubMed, Embase, Cochrane Library, and Web of Science for studies published before March 2024. The Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) was used to assess the methodological quality of the included studies, random-effects models were used to calculate sensitivity and specificity, I2 values were used for heterogeneity measurements, and subgroup analyses were carried out to detect potential sources of heterogeneity.

RESULTS

A total of 1699 patients from 24 studies were included. For machine learning models in predicting pCR to nCRT, the meta-analysis calculated a pooled area under the curve (AUC) of 0.91 (95% CI, 0.88-0.93), pooled sensitivity of 0.83 (95% CI, 0.74-0.89), and pooled specificity of 0.86 (95% CI, 0.80-0.91). We investigated 6 studies that mainly contributed to heterogeneity. After performing meta-analysis again excluding these 6 studies, the heterogeneity was significantly reduced. In subgroup analysis, the pooled AUC of the deep-learning model was 0.93 and 0.89 for the traditional statistical model; the pooled AUC of studies that used diffusion-weighted imaging (DWI) was 0.90 and 0.92 in studies that did not use DWI; the pooled AUC of studies conducted in China was 0.93, and was 0.83 in studies conducted in other countries.

CONCLUSIONS

This systematic study showed that machine learning has promising potential in predicting pCR to nCRT in patients with locally advanced rectal cancer. Compared to traditional machine learning models, although deep-learning-based studies are less predominant and more heterogeneous, they are able to obtain higher AUC.

ADVANCES IN KNOWLEDGE

Compared to traditional machine learning models, deep-learning-based studies are able to obtain higher AUC, although they are less predominant and more heterogeneous. Together with clinical information, machine learning-based models may bring us closer towards precision medicine.

An automated deep learning pipeline for EMVI classification and response prediction of rectal cancer using baseline MRI: a multi-centre study

The classification of extramural vascular invasion status using baseline magnetic resonance imaging in rectal cancer has gained significant attention as it is an important prognostic marker. Also, the accurate prediction of patients achieving complete response with primary staging MRI assists clinicians in determining subsequent treatment plans. Most studies utilised radiomics-based methods, requiring manually annotated segmentation and handcrafted features, which tend to generalise poorly. We retrospectively collected 509 patients from 9 centres, and proposed a fully automated pipeline for EMVI status classification and CR prediction with diffusion weighted imaging and T2-weighted imaging. We applied nnUNet, a self-configuring deep learning model, for tumour segmentation and employed learned multiple-level image features to train classification models, named MLNet. This ensures a more comprehensive representation of the tumour features, in terms of both fine-grained detail and global context. On external validation, MLNet, yielding similar AUCs as internal validation, outperformed 3D ResNet10, a deep neural network with ten layers designed for analysing spatiotemporal data, in both CR and EMVI tasks. For CR prediction, MLNet showed better results than the current state-of-the-art model using imaging and clinical features in the same external cohort. Our study demonstrated that incorporating multi-level image representations learned by a deep learning based tumour segmentation model on primary MRI improves the results of EMVI classification and CR prediction with good generalisation to external data. We observed variations in the contributions of individual feature maps to different classification tasks. This pipeline has the potential to be applied in clinical settings, particularly for EMVI classification.

Artificial Intelligence Applications in the Treatment of Colorectal Cancer: A Narrative Review

Colorectal cancer is the third most prevalent cancer worldwide, and its treatment has been a demanding clinical problem. Beyond traditional surgical therapy and chemotherapy, newly revealed molecular mechanisms diversify therapeutic approaches for colorectal cancer. However, the selection of personalized treatment among multiple treatment options has become another challenge in the era of precision medicine. Artificial intelligence has recently been increasingly investigated in the treatment of colorectal cancer. This narrative review mainly discusses the applications of artificial intelligence in the treatment of colorectal cancer patients. A comprehensive literature search was conducted in MEDLINE, EMBASE, and Web of Science to identify relevant papers, resulting in 49 articles being included. The results showed that, based on different categories of data, artificial intelligence can predict treatment outcomes and essential guidance information of traditional and novel therapies, thus enabling individualized treatment strategy selection for colorectal cancer patients. Some frequently implemented machine learning algorithms and deep learning frameworks have also been employed for long-term prognosis prediction in patients with colorectal cancer. Overall, artificial intelligence shows encouraging results in treatment strategy selection and prognosis evaluation for colorectal cancer patients.

Multi-task deep learning-based radiomic nomogram for prognostic prediction in locoregionally advanced nasopharyngeal carcinoma

PURPOSE

Prognostic prediction is crucial to guide individual treatment for locoregionally advanced nasopharyngeal carcinoma (LA-NPC) patients. Recently, multi-task deep learning was explored for joint prognostic prediction and tumor segmentation in various cancers, resulting in promising performance. This study aims to evaluate the clinical value of multi-task deep learning for prognostic prediction in LA-NPC patients.

METHODS

A total of 886 LA-NPC patients acquired from two medical centers were enrolled including clinical data, [18F]FDG PET/CT images, and follow-up of progression-free survival (PFS). We adopted a deep multi-task survival model (DeepMTS) to jointly perform prognostic prediction (DeepMTS-Score) and tumor segmentation from FDG-PET/CT images. The DeepMTS-derived segmentation masks were leveraged to extract handcrafted radiomics features, which were also used for prognostic prediction (AutoRadio-Score). Finally, we developed a multi-task deep learning-based radiomic (MTDLR) nomogram by integrating DeepMTS-Score, AutoRadio-Score, and clinical data. Harrell's concordance indices (C-index) and time-independent receiver operating characteristic (ROC) analysis were used to evaluate the discriminative ability of the proposed MTDLR nomogram. For patient stratification, the PFS rates of high- and low-risk patients were calculated using Kaplan-Meier method and compared with the observed PFS probability.

RESULTS

Our MTDLR nomogram achieved C-index of 0.818 (95% confidence interval (CI): 0.785-0.851), 0.752 (95% CI: 0.638-0.865), and 0.717 (95% CI: 0.641-0.793) and area under curve (AUC) of 0.859 (95% CI: 0.822-0.895), 0.769 (95% CI: 0.642-0.896), and 0.730 (95% CI: 0.634-0.826) in the training, internal validation, and external validation cohorts, which showed a statistically significant improvement over conventional radiomic nomograms. Our nomogram also divided patients into significantly different high- and low-risk groups.

CONCLUSION

Our study demonstrated that MTDLR nomogram can perform reliable and accurate prognostic prediction in LA-NPC patients, and also enabled better patient stratification, which could facilitate personalized treatment planning.

Clinical evaluation on automatic segmentation results of convolutional neural networks in rectal cancer radiotherapy

Purpose

Image segmentation can be time-consuming and lacks consistency between different oncologists, which is essential in conformal radiotherapy techniques. We aimed to evaluate automatic delineation results generated by convolutional neural networks (CNNs) from geometry and dosimetry perspectives and explore the reliability of these segmentation tools in rectal cancer.

Methods

Forty-seven rectal cancer cases treated from February 2018 to April 2019 were randomly collected retrospectively in our cancer center. The oncologists delineated regions of interest (ROIs) on planning CT images as the ground truth, including clinical target volume (CTV), bladder, small intestine, and femoral heads. The corresponding automatic segmentation results were generated by DeepLabv3+ and ResUNet, and we also used Atlas-Based Autosegmentation (ABAS) software for comparison. The geometry evaluation was carried out using the volumetric Dice similarity coefficient (DSC) and surface DSC, and critical dose parameters were assessed based on replanning optimized by clinically approved or automatically generated CTVs and organs at risk (OARs), i.e., the Planref and Plantest. Pearson test was used to explore the correlation between geometric metrics and dose parameters.

Results

In geometric evaluation, DeepLabv3+ performed better in DCS metrics for the CTV (volumetric DSC, mean = 0.96, P< 0.01; surface DSC, mean = 0.78, P< 0.01) and small intestine (volumetric DSC, mean = 0.91, P< 0.01; surface DSC, mean = 0.62, P< 0.01), ResUNet had advantages in volumetric DSC of the bladder (mean = 0.97, P< 0.05). For critical dose parameters analysis between Planref and Plantest, there was a significant difference for target volumes (P< 0.01), and no significant difference was found for the ResUNet-generated small intestine (P > 0.05). For the correlation test, a negative correlation was found between DSC metrics (volumetric, surface DSC) and dosimetric parameters (δD95, δD95, HI, CI) for target volumes (P< 0.05), and no significant correlation was found for most tests of OARs (P > 0.05).

Conclusions

CNNs show remarkable repeatability and time-saving in automatic segmentation, and their accuracy also has a certain potential in clinical practice. Meanwhile, clinical aspects, such as dose distribution, may need to be considered when comparing the performance of auto-segmentation methods.

Radiomics from Mesorectal Blood Vessels and Lymph Nodes: A Novel Prognostic Predictor for Rectal Cancer with Neoadjuvant Therapy

The objective of our study is to investigate the predictive value of various combinations of radiomic features from intratumoral and different peritumoral regions of interest (ROIs) for achieving a good pathological response (pGR) following neoadjuvant chemoradiotherapy (nCRT) in patients with locally advanced rectal cancer (LARC). This retrospective study was conducted using data from LARC patients who underwent nCRT between 2013 and 2021. Patients were divided into training and validation cohorts at a ratio of 4:1. Intratumoral ROIs (ROI_ITU) were segmented on T2-weighted imaging, while peritumoral ROIs were segmented using two methods: ROI_{PTU_2mm}, ROI_{PTU_4mm}, and ROI_{PTU_6mm}, obtained by dilating the boundary of ROI_ITU by 2 mm, 4 mm, and 6 mm, respectively; and ROI_{MR_F} and ROI_{MR_BVLN}, obtained by separating the fat and blood vessels + lymph nodes in the mesorectum. After feature extraction and selection, 12 logistic regression models were established using radiomics features derived from different ROIs or ROI combinations, and five-fold cross-validation was performed. The average area under the receiver operating characteristic curve (AUC) was used to evaluate the performance of the models. The study included 209 patients, consisting of 118 pGR and 91 non-pGR patients. The model that integrated ROI_ITU and ROI_{MR_BVLN} features demonstrated the highest predictive ability, with an AUC (95% confidence interval) of 0.936 (0.904-0.972) in the training cohort and 0.859 (0.745-0.974) in the validation cohort. This model outperformed models that utilized ROI_ITU alone (AUC = 0.779), ROI_{MR_BVLN} alone (AUC = 0.758), and other models. The radscore derived from the optimal model can predict the treatment response and prognosis after nCRT. Our findings validated that the integration of intratumoral and peritumoral radiomic features, especially those associated with mesorectal blood vessels and lymph nodes, serves as a potent predictor of pGR to nCRT in patients with LARC. Pending further corroboration in future research, these insights could provide novel imaging markers for refining therapeutic strategies.

Normalization Strategies in Multi-Center Radiomics Abdominal MRI: Systematic Review and Meta-Analyses

Goal: Artificial intelligence applied to medical image analysis has been extensively used to develop non-invasive diagnostic and prognostic signatures. However, these imaging biomarkers should be largely validated on multi-center datasets to prove their robustness before they can be introduced into clinical practice. The main challenge is represented by the great and unavoidable image variability which is usually addressed using different pre-processing techniques including spatial, intensity and feature normalization. The purpose of this study is to systematically summarize normalization methods and to evaluate their correlation with the radiomics model performances through meta-analyses. This review is carried out according to the PRISMA statement: 4777 papers were collected, but only 74 were included. Two meta-analyses were carried out according to two clinical aims: characterization and prediction of response. Findings of this review demonstrated that there are some commonly used normalization approaches, but not a commonly agreed pipeline that can allow to improve performance and to bridge the gap between bench and bedside.

Independent Validation of a Deep Learning nnU-Net Tool for Neuroblastoma Detection and Segmentation in MR Images

OBJECTIVES

To externally validate and assess the accuracy of a previously trained fully automatic nnU-Net CNN algorithm to identify and segment primary neuroblastoma tumors in MR images in a large children cohort.

METHODS

An international multicenter, multivendor imaging repository of patients with neuroblastic tumors was used to validate the performance of a trained Machine Learning (ML) tool to identify and delineate primary neuroblastoma tumors. The dataset was heterogeneous and completely independent from the one used to train and tune the model, consisting of 300 children with neuroblastic tumors having 535 MR T2-weighted sequences (486 sequences at diagnosis and 49 after finalization of the first phase of chemotherapy). The automatic segmentation algorithm was based on a nnU-Net architecture developed within the PRIMAGE project. For comparison, the segmentation masks were manually edited by an expert radiologist, and the time for the manual editing was recorded. Different overlaps and spatial metrics were calculated to compare both masks.

RESULTS

The median Dice Similarity Coefficient (DSC) was high 0.997; 0.944-1.000 (median; Q1-Q3). In 18 MR sequences (6%), the net was not able neither to identify nor segment the tumor. No differences were found regarding the MR magnetic field, type of T2 sequence, or tumor location. No significant differences in the performance of the net were found in patients with an MR performed after chemotherapy. The time for visual inspection of the generated masks was 7.9 ± 7.5 (mean ± Standard Deviation (SD)) seconds. Those cases where manual editing was needed (136 masks) required 124 ± 120 s.

CONCLUSIONS

The automatic CNN was able to locate and segment the primary tumor on the T2-weighted images in 94% of cases. There was an extremely high agreement between the automatic tool and the manually edited masks. This is the first study to validate an automatic segmentation model for neuroblastic tumor identification and segmentation with body MR images. The semi-automatic approach with minor manual editing of the deep learning segmentation increases the radiologist's confidence in the solution with a minor workload for the radiologist.

Radiomics Approaches for the Prediction of Pathological Complete Response after Neoadjuvant Treatment in Locally Advanced Rectal Cancer: Ready for Prime Time?

In recent years, neoadjuvant therapy of locally advanced rectal cancer has seen tremendous modifications. Adding neoadjuvant chemotherapy before or after chemoradiotherapy significantly increases loco-regional disease-free survival, negative surgical margin rates, and complete response rates. The higher complete rate is particularly clinically meaningful given the possibility of organ preservation in this specific sub-population, without compromising overall survival. However, all locally advanced rectal cancer most likely does not benefit from total neoadjuvant therapy (TNT), but experiences higher toxicity rates. Diagnosis of complete response after neoadjuvant therapy is a real challenge, with a risk of false negatives and possible under-treatment. These new therapeutic approaches thus raise the need for better selection tools, enabling a personalized therapeutic approach for each patient. These tools mostly focus on the prediction of the pathological complete response given the clinical impact. In this article, we review the place of different biomarkers (clinical, biological, genomics, transcriptomics, proteomics, and radiomics) as well as their clinical implementation and discuss the most recent trends for future steps in prediction modeling in patients with locally advanced rectal cancer.

MRI-based pre-Radiomics and delta-Radiomics models accurately predict the post-treatment response of rectal adenocarcinoma to neoadjuvant chemoradiotherapy

Objectives

To develop and validate magnetic resonance imaging (MRI)-based pre-Radiomics and delta-Radiomics models for predicting the treatment response of local advanced rectal cancer (LARC) to neoadjuvant chemoradiotherapy (NCRT).

Methods

Between October 2017 and August 2022, 105 LARC NCRT-naïve patients were enrolled in this study. After careful evaluation, data for 84 patients that met the inclusion criteria were used to develop and validate the NCRT response models. All patients received NCRT, and the post-treatment response was evaluated by pathological assessment. We manual segmented the volume of tumors and 105 radiomics features were extracted from three-dimensional MRIs. Then, the eXtreme Gradient Boosting algorithm was implemented for evaluating and incorporating important tumor features. The predictive performance of MRI sequences and Synthetic Minority Oversampling Technique (SMOTE) for NCRT response were compared. Finally, the optimal pre-Radiomics and delta-Radiomics models were established respectively. The predictive performance of the radionics model was confirmed using 5-fold cross-validation, 10-fold cross-validation, leave-one-out validation, and independent validation. The predictive accuracy of the model was based on the area under the receiver operator characteristic (ROC) curve (AUC).

Results

There was no significant difference in clinical factors between patients with good and poor reactions. Integrating different MRI modes and the SMOTE method improved the performance of the radiomics model. The pre-Radiomics model (train AUC: 0.93 ± 0.06; test AUC: 0.79) and delta-Radiomcis model (train AUC: 0.96 ± 0.03; test AUC: 0.83) all have high NCRT response prediction performance by LARC. Overall, the delta-Radiomics model was superior to the pre-Radiomics model.

Conclusion

MRI-based pre-Radiomics model and delta-Radiomics model all have good potential to predict the post-treatment response of LARC to NCRT. Delta-Radiomics analysis has a huge potential for clinical application in facilitating the provision of personalized therapy.

Robustness of Radiomics in Pre-Surgical Computer Tomography of Non-Small-Cell Lung Cancer

Background: Radiomic features are increasingly used in CT of NSCLC. However, their robustness with respect to segmentation variability has not yet been demonstrated. The aim of this study was to assess radiomic features agreement across three kinds of segmentation. Methods: We retrospectively included 48 patients suffering from NSCLC who underwent pre-surgery CT. Two expert radiologists in consensus manually delineated three 3D-ROIs on each patient. To assess robustness for each feature, the intra-class correlation coefficient (ICC) across segmentations was evaluated. The ‘sensitivity’ of ICC upon some parameters affecting features computation (such as bin-width for first-order features and pixel-distances for second-order features) was also evaluated. Moreover, an assessment with respect to interpolator and isotropic resolution was also performed. Results: Our results indicate that ‘shape’ features tend to have excellent agreement (ICC > 0.9) across segmentations; moreover, they have approximately zero sensitivity to other parameters. ‘First-order’ features are in general sensitive to parameters variation; however, a few of them showed excellent agreement and low sensitivity (below 0.1) with respect to bin-width and pixel-distance. Similarly, a few second-order features showed excellent agreement and low sensitivity. Conclusions: Our results suggest that a limited number of radiomic features can achieve a high level of reproducibility in CT of NSCLC.

Enhancing the stability of CT radiomics across different volume of interest sizes using parametric feature maps: a phantom study

BACKGROUND

In radiomics studies, differences in the volume of interest (VOI) are often inevitable and may confound the extracted features. We aimed to correct this confounding effect of VOI variability by applying parametric maps with a fixed voxel size.

METHODS

Ten scans of a cup filled with sodium chloride solution were scanned using a multislice computed tomography (CT) unit. Sphere-shaped VOIs with different diameters (4, 8, or 16 mm) were drawn centrally into the phantom. A total of 93 features were extracted conventionally from the original images using PyRadiomics. Using a self-designed and pretested software tool, parametric maps for the same 93 features with a fixed voxel size of 4 mm3 were created. To retrieve the feature values from the maps, VOIs were copied from the original images to preserve the position. Differences in feature quantities between the VOI sizes were tested with the Mann-Whitney U-test and agreement with overall concordance correlation coefficients (OCCC).

RESULTS

Fifty-five conventionally extracted features were significantly different between the VOI sizes, and none of the features showed excellent agreement in terms of OCCCs. When read from the parametric maps, only 8 features showed significant differences, and 3 features showed an excellent OCCC (≥ 0.85). The OCCCs for 89 features substantially increased using the parametric maps.

CONCLUSIONS

This phantom study shows that converting CT images into parametric maps resolves the confounding effect of VOI variability and increases feature reproducibility across VOI sizes.