Validation of a deep learning segmentation algorithm to quantify the skeletal muscle index and sarcopenia in metastatic renal carcinoma

Predicting malnutrition in gastric cancer patients using computed tomography(CT) deep learning features and clinical data

OBJECTIVE

The aim of this study is using clinical factors and non-enhanced computed tomography (CT) deep features of the psoas muscles at third lumbar vertebral (L3) level to construct a model to predict malnutrition in gastric cancer before surgery, and to provide a new nutritional status assessment and survival assessment tool for gastric cancer patients.

METHODS

A retrospective analysis of 312 patients of gastric cancer were divided into malnutrition group and normal group based on Nutrition Risk Screening 2002(NRS-2002). 312 regions of interest (ROI) of the psoas muscles at L3 level of non-enhanced CT were delineated. Deep learning (DL) features were extracted from the ROI using a deep migration model and were screened by principal component analysis (PCA) and least-squares operator (LASSO). The clinical predictors included Body Mass Index (BMI), lymphocyte and albumin. Both deep learning model (including deep learning features) and mixed model (including selected deep learning features and selected clinical predictors) were constructed by 11 classifiers. The model was evaluated and selected by calculating receiver operating characteristic (ROC), area under curve (AUC), accuracy, sensitivity and specificity, calibration curve and decision curve analysis (DCA). The Cohen's Kappa coefficient (κ) was using to compare the diagnostic agreement for malnutrition between the mixed model and the GLIM in gastric cancer patients.

RESULT

The results of logistics multivariate analysis showed that BMI [OR = 0.569 (95% CI 0.491-0.660)], lymphocyte [OR = 0.638 (95% CI 0.408-0.998)], and albumin [OR = 0.924 (95% CI 0.859-0.994)] were clinically independent malnutrition of gastric cancer predictor(P < 0.05). Among the 11 classifiers, the Multilayer Perceptron (MLP)were selected as the best classifier. The AUC of the training and test sets for deep learning model were 0.806 (95% CI 0.7485-0.8635) and 0.769 (95% CI 0.673-0.863) and with accuracies were 0.734 and 0.766, respectively. The AUC of the training and test sets for the mixed model were 0.909 (95% CI 0.869-0.948) and 0.857 (95% CI 0.782-0.931) and with accuracies of 0.845 and 0.861, respectively. The DCA confirmed the clinical benefit of the both models. The Cohen's Kappa coefficient (κ) was 0.647 (P < 0.001). Diagnostic agreement for malnutrition between the mixed model and GLIM criteria was good. The mixed model was used to calculate the predicted probability of malnutrition in gastric cancer patients, which was divided into high-risk and low-risk groups by median, and the survival analysis showed that the overall survival time of the high-risk group was significantly lower than that of the low-risk group (P = 0.005).

CONCLUSION

Deep learning based on mixed model may be a potential tool for predicting malnutrition in gastric cancer patients.

Best imaging signs identified by radiomics could outperform the model: application to differentiating lung carcinoid tumors from atypical hamartomas

OBJECTIVES

Lung carcinoids and atypical hamartomas may be difficult to differentiate but require different treatment. The aim was to differentiate these tumors using contrast-enhanced CT semantic and radiomics criteria.

METHODS

Between November 2009 and June 2020, consecutives patient operated for hamartomas or carcinoids with contrast-enhanced chest-CT were retrospectively reviewed. Semantic criteria were recorded and radiomics features were extracted from 3D segmentations using Pyradiomics. Reproducible and non-redundant radiomics features were used to training a random forest algorithm with cross-validation. A validation-set from another institution was used to evaluate of the radiomics signature, the 3D 'median' attenuation feature (3D-median) alone and the mean value from 2D-ROIs.

RESULTS

Seventy-three patients (median 58 years [43‒70]) were analyzed (16 hamartomas; 57 carcinoids). The radiomics signature predicted hamartomas vs carcinoids on the external dataset (22 hamartomas; 32 carcinoids) with an AUC = 0.76. The 3D-median was the most important in the model. Density thresholds < 10 HU to predict hamartoma and > 60 HU to predict carcinoids were chosen for their high specificity > 0.90. On the external dataset, sensitivity and specificity of the 3D-median and 2D-ROIs were, respectively, 0.23, 1.00 and 0.13, 1.00 < 10 HU; 0.63, 0.95 and 0.69, 0.91 > 60 HU. The 3D-median was more reproducible than 2D-ROIs (ICC = 0.97 95% CI [0.95‒0.99]; bias: 3 ± 7 HU limits of agreement (LoA) [- 10‒16] vs. ICC = 0.90 95% CI [0.85‒0.94]; bias: - 0.7 ± 21 HU LoA [- 4‒40], respectively).

CONCLUSIONS

A radiomics signature can distinguish hamartomas from carcinoids with an AUC = 0.76. Median density < 10 HU and > 60 HU on 3D or 2D-ROIs may be useful in clinical practice to diagnose these tumors with confidence, but 3D is more reproducible.

CRITICAL RELEVANCE STATEMENT

Radiomic features help to identify the most discriminating imaging signs using random forest. 'Median' attenuation value (Hounsfield units), extracted from 3D-segmentations on contrast-enhanced chest-CTs, could distinguish carcinoids from atypical hamartomas (AUC = 0.85), was reproducible (ICC = 0.97), and generalized to an external dataset.

KEY POINTS

• 3D-'Median' was the best feature to differentiate carcinoids from atypical hamartomas (AUC = 0.85). • 3D-'Median' feature is reproducible (ICC = 0.97) and was generalized to an external dataset. • Radiomics signature from 3D-segmentations differentiated carcinoids from atypical hamartomas with an AUC = 0.76. • 2D-ROI value reached similar performance to 3D-'median' but was less reproducible (ICC = 0.90).

Comparative performances of machine learning algorithms in radiomics and impacting factors

There are no current recommendations on which machine learning (ML) algorithms should be used in radiomics. The objective was to compare performances of ML algorithms in radiomics when applied to different clinical questions to determine whether some strategies could give the best and most stable performances regardless of datasets. This study compares the performances of nine feature selection algorithms combined with fourteen binary classification algorithms on ten datasets. These datasets included radiomics features and clinical diagnosis for binary clinical classifications including COVID-19 pneumonia or sarcopenia on CT, head and neck, orbital or uterine lesions on MRI. For each dataset, a train-test split was created. Each of the 126 (9 × 14) combinations of feature selection algorithms and classification algorithms was trained and tuned using a ten-fold cross validation, then AUC was computed. This procedure was repeated three times per dataset. Best overall performances were obtained with JMI and JMIM as feature selection algorithms and random forest and linear regression models as classification algorithms. The choice of the classification algorithm was the factor explaining most of the performance variation (10% of total variance). The choice of the feature selection algorithm explained only 2% of variation, while the train-test split explained 9%.

Standardization of Body Composition Status in Patients with Advanced Urothelial Tumors: The Role of a CT-Based AI-Powered Software for the Assessment of Sarcopenia and Patient Outcome Correlation

BACKGROUND

Sarcopenia is a well know prognostic factor in oncology, influencing patients' quality of life and survival. We aimed to investigate the role of sarcopenia, assessed by a Computed Tomography (CT)-based artificial intelligence (AI)-powered-software, as a predictor of objective clinical benefit in advanced urothelial tumors and its correlations with oncological outcomes.

METHODS

We retrospectively searched patients with advanced urothelial tumors, treated with systemic platinum-based chemotherapy and an available total body CT, performed before and after therapy. An AI-powered software was applied to CT to obtain the Skeletal Muscle Index (SMI-L3), derived from the area of the psoas, long spine, and abdominal muscles, at the level of L3 on CT axial images. Logistic and Cox-regression modeling was implemented to explore the association of sarcopenic status and anthropometric features to the clinical benefit rate and survival endpoints.

RESULTS

97 patients were included, 66 with bladder cancer and 31 with upper-tract urothelial carcinoma. Clinical benefit outcomes showed a linear positive association with all the observed body composition variables variations. The chances of not experiencing disease progression were positively associated with ∆_SMI-L3, ∆_psoas, and ∆_long spine muscle when they ranged from ~10-20% up to ~45-55%. Greater survival chances were matched by patients achieving a wider ∆_SMI-L3, ∆_abdominal and ∆_long spine muscle.

CONCLUSIONS

A CT-based AI-powered software body composition and sarcopenia analysis provide prognostic assessments for objective clinical benefits and oncological outcomes.

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.

Application of artificial intelligence to imaging interpretations in the musculoskeletal area: Where are we? Where are we going?

The interest of researchers, clinicians and radiologists, in artificial intelligence (AI) continues to grow. Deep learning is a subset of machine learning, in which the computer algorithm itself can determine the optimal imaging features to answer a clinical question. Convolutional neural networks are the most common architecture for performing deep learning on medical images. The various musculoskeletal applications of deep learning are the detection of abnormalities on X-rays or cross-sectional images (CT, MRI), for example the detection of fractures, meniscal tears, anterior cruciate ligament tears, degenerative lesions of the spine, bone metastases, classification of e.g., dural sac stenosis, degeneration of intervertebral discs, assessment of skeletal age, and segmentation, for example of cartilage. Software developments are already impacting the daily practice of orthopedic imaging by automatically detecting fractures on radiographs. Improving image acquisition protocols, improving the quality of low-dose CT images, reducing acquisition times in MRI, or improving MR image resolution is possible through deep learning. Deep learning offers an automated way to offload time-consuming manual processes and improve practitioner performance. This article reviews the current state of AI in musculoskeletal imaging.