Machine Learning Approach to Differentiation of Peripheral Schwannomas and Neurofibromas: A Multi-Center Study

Abnormal Brachial Plexus Differentiation from Routine Magnetic Resonance Imaging: An AI-based Approach

Automatic abnormality identification of brachial plexus (BP) from normal magnetic resonance imaging to localize and identify a neurologic injury in clinical practice (MRI) is still a novel topic in brachial plexopathy. This study developed and evaluated an approach to differentiate abnormal BP with artificial intelligence (AI) over three commonly used MRI sequences, i.e. T1, FLUID sensitive and post-gadolinium sequences. A BP dataset was collected by radiological experts and a semi-supervised artificial intelligence method was used to segment the BP (based on nnU-net). Hereafter, a radiomics method was utilized to extract 107 shape and texture features from these ROIs. From various machine learning methods, we selected six widely recognized classifiers for training our Brachial plexus (BP) models and assessing their efficacy. To optimize these models, we introduced a dynamic feature selection approach aimed at discarding redundant and less informative features. Our experimental findings demonstrated that, in the context of identifying abnormal BP cases, shape features displayed heightened sensitivity compared to texture features. Notably, both the Logistic classifier and Bagging classifier outperformed other methods in our study. These evaluations illuminated the exceptional performance of our model trained on FLUID-sensitive sequences, which notably exceeded the results of both T1 and post-gadolinium sequences. Crucially, our analysis highlighted that both its classification accuracies and AUC score (area under the curve of receiver operating characteristics) over FLUID-sensitive sequence exceeded 90%. This outcome served as a robust experimental validation, affirming the substantial potential and strong feasibility of integrating AI into clinical practice.

MRI features of benign peripheral nerve sheath tumors: how do sporadic and syndromic tumors differ?

OBJECTIVES

To compare MRI features of sporadic and neurofibromatosis syndrome-related localized schwannomas and neurofibromas.

METHODS

In this retrospective study, our pathology database was searched for "neurofibroma" or "schwannoma" from 2014 to 2019. Exclusion criteria were lack of available MRI and intradural or plexiform tumors. Qualitative and quantitative anatomic (location, size, relationship to nerve, signal, muscle denervation) and functional (arterial enhancement, apparent diffusion-weighted coefficient) MRI features of sporadic and syndrome-related tumors were compared. Statistical significance was assumed for p < 0.05.

RESULTS

A total of 80 patients with 64 schwannomas (sporadic: 42 (65.6%) v. syndrome-related: 22 (34.4%)) and 19 neurofibromas (sporadic: 7 (36.8%) v. syndrome-related: 12 (41.7%)) were included. Only signal heterogeneity (T2W p=0.001, post-contrast p=0.03) and a diffused-weighted imaging target sign (p=0.04) were more frequent with schwannomas than neurofibromas. Sporadic schwannomas were similar in size to syndrome-related schwannomas (2.9±1.2cm vs. 3.7±3.2 cm, p = 0.6), but with greater heterogeneity (T2W p = 0.02, post-contrast p = 0.01). Sporadic neurofibromas were larger (4.6±1.5cm vs. 3.4±2.4 cm, p = 0.03) than syndrome-related neurofibromas, also with greater heterogeneity (T2W p=0.03, post-contrast p=0.04). Additional tumors along an affected nerve were only observed with syndrome-related tumors). There was no difference in apparent diffusion coefficient values or presence of early perfusion between sporadic and syndrome-related tumors (p > 0.05).

CONCLUSIONS

Although syndrome-related and sporadic schwannomas and neurofibromas overlap in their anatomic, diffusion and perfusion features, signal heterogeneity and presence of multiple lesions along a nerve are differentiating characteristics of syndrome-related tumors.

Establishment and Optimization of Radiomics Algorithms for Prediction of KRAS Gene Mutation by Integration of NSCLC Gene Mutation Mutual Exclusion Information

Purpose: To assess the significance of mutation mutual exclusion information in the optimization of radiomics algorithms for predicting gene mutation. Methods: We retrospectively analyzed 258 non-small cell lung cancer (NSCLC) patients. Patients were randomly divided into training (n = 180) and validation (n = 78) cohorts. Based on radiomics features, radiomics score (RS) models were developed for predicting KRAS proto-oncogene mutations. Furthermore, a composite model combining mixedRS and epidermal growth factor receptor (EGFR) mutation status was developed. Results: Compared with CT model, the PET/CT radiomics score model exhibited higher AUC for predicting KRAS mutations (0.834 vs. 0.770). By integrating EGFR mutation information into the PET/CT RS model, the AUC, sensitivity, specificity, and accuracy for predicting KRAS mutations were all elevated in the validation cohort (0.921, 0.949, 0.872, 0.910 vs. 0.834, 0.923, 0.641, 0.782). By adding EGFR exclusive mutation information, the composite model corrected 64.3% false positive cases produced by the PET/CT RS model in the validation cohort. Conclusion: Integrating EGFR mutation status has potential utility for the optimization of radiomics models for prediction of KRAS gene mutations. This method may be used when repeated biopsies would carry unacceptable risks for the patient.

Classification of Hepatocellular Carcinoma and Intrahepatic Cholangiocarcinoma Based on Radiomic Analysis

Introduction

Considering the narrow window of surgery, early diagnosis of liver cancer is still a fundamental issue to explore. Hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICCA) are considered as two different types of liver cancer because of their distinct pathogenesis, pathological features, prognosis, and responses to adjuvant therapies. Qualitative analysis of image is not enough to make a discrimination of liver cancer, especially early-stage HCC or ICCA.

Methods

This retrospective study developed a radiomic-based model in a training cohort of 122 patients. Radiomic features were extracted from computed tomography (CT) scans. Feature selection was operated with the least absolute shrinkage and operator (LASSO) logistic method. The support vector machine (SVM) was selected to build a model. An internal validation was conducted in 89 patients.

Results

In the training set, the AUC of the evaluation of the radiomics was 0.855 higher than for radiologists at 0.689. In the valuation cohorts, the AUC of the evaluation was 0.847 and the validation was 0.659, which indicated that the established model has a significantly better performance in distinguishing the HCC from ICCA.

Conclusion

We developed a radiomic diagnosis model based on CT image that can quickly distinguish HCC from ICCA, which may facilitate the differential diagnosis of HCC and ICCA in the future.