Least square-support vector machine based brain tumor classification system with multi model texture features

Differentiation of Early Sacroiliitis Using Machine-Learning- Supported Texture Analysis

Objectives: We wished to compare the diagnostic performance of texture analysis (TA) against that of a visual qualitative assessment in identifying early sacroiliitis (nr-axSpA). Methods: A total of 92 participants were retrospectively included at our university hospital institution, comprising 30 controls and 62 patients with axSpA, including 32 with nr-axSpA and 30 with r-axSpA, who underwent MR examination of the sacroiliac joints. MRI at 3T of the lumbar spine and the sacroiliac joint was performed using oblique T1-weighted (W), fluid-sensitive, fat-saturated (Fs) T2WI images. The modified New York criteria for AS were used. Patients were classified into the nr-axSpA group if their digital radiography (DR) and/or CT results within 7 days from the MR examination showed a DR and/or CT grade < 2 for the bilateral sacroiliac joints or a DR and/or CT grade < 3 for the unilateral sacroiliac joint. Patients were classified into the r-axSpA group if their DR and/or CT grade was 2 to 3 for the bilateral sacroiliac joints or their DR and/or CT grade was 3 for the unilateral sacroiliac joint. Patients were considered to have a confirmed diagnosis if their DR or CT grade was 4 for the sacroiliac joints and were thereby excluded. A control group of healthy individuals matched in terms of age and sex to the patients was included in this study. First, two readers independently qualitatively scored the oblique coronal T1WI and FsT2WI non-enhanced sacroiliac joint images. The diagnostic efficacies of the two readers were judged and compared using an assigned Likert score, conducting a Kappa consistency test of the diagnostic results between two readers. Texture analysis models (the T1WI-TA model and the FsT2WI-TA model) were constructed through feature extraction and feature screening. The qualitative and quantitative results were evaluated for their diagnostic performance and compared against a clinical reference standard. Results: The qualitative scores of the two readers could significantly distinguish between the healthy controls and the nr-axSpA group and the nr-axSpA and r-axSpA groups (both p < 0.05). Both TA models could significantly distinguish between the healthy controls and the nr-axSpA group and the nr-axSpA group and the r-axSpA group (both p < 0.05). There was no significant difference in the differential diagnoses of the two TA models between the healthy controls and the nr-axSpA group (AUC: 0.934 vs. 0.976; p = 0.1838) and between the nr-axSpA and r-axSpA groups (AUC: 0.917 vs. 0.848; p = 0.2592). In terms of distinguishing between the healthy control and nr-axSpA groups, both the TA models were superior to the qualitative scores of the two readers (all p < 0.05). In terms of distinguishing between the nr-axSpA and r-axSpA groups, the T1WI-TA model was superior to the qualitative scores of the two readers (p = 0.023 and p = 0.007), whereas there was no significant difference between the fsT2WI-TA model and the qualitative scores of the two readers (p = 0.134 and p = 0.065). Conclusions: Based on MR imaging, the T1WI-TA and fsT2WI-TA models were highly effective for the early diagnosis of sacroiliac joint arthritis. The T1WI-TA model significantly improved the early diagnostic efficacy for sacroiliac arthritis compared to that of the qualitative scores of the readers, while the efficacy of the fsT2WI-TA model was comparable to that of the readers.

Computer-aided diagnosis system for grading brain tumor using histopathology images based on color and texture features

BACKGROUND

Cancer pathology shows disease development and associated molecular features. It provides extensive phenotypic information that is cancer-predictive and has potential implications for planning treatment. Based on the exceptional performance of computational approaches in the field of digital pathogenic, the use of rich phenotypic information in digital pathology images has enabled us to identify low-level gliomas (LGG) from high-grade gliomas (HGG). Because the differences between the textures are so slight, utilizing just one feature or a small number of features produces poor categorization results.

METHODS

In this work, multiple feature extraction methods that can extract distinct features from the texture of histopathology image data are used to compare the classification outcomes. The successful feature extraction algorithms GLCM, LBP, multi-LBGLCM, GLRLM, color moment features, and RSHD have been chosen in this paper. LBP and GLCM algorithms are combined to create LBGLCM. The LBGLCM feature extraction approach is extended in this study to multiple scales using an image pyramid, which is defined by sampling the image both in space and scale. The preprocessing stage is first used to enhance the contrast of the images and remove noise and illumination effects. The feature extraction stage is then carried out to extract several important features (texture and color) from histopathology images. Third, the feature fusion and reduction step is put into practice to decrease the number of features that are processed, reducing the computation time of the suggested system. The classification stage is created at the end to categorize various brain cancer grades. We performed our analysis on the 821 whole-slide pathology images from glioma patients in the Cancer Genome Atlas (TCGA) dataset. Two types of brain cancer are included in the dataset: GBM and LGG (grades II and III). 506 GBM images and 315 LGG images are included in our analysis, guaranteeing representation of various tumor grades and histopathological features.

RESULTS

The fusion of textural and color characteristics was validated in the glioma patients using the 10-fold cross-validation technique with an accuracy equals to 95.8%, sensitivity equals to 96.4%, DSC equals to 96.7%, and specificity equals to 97.1%. The combination of the color and texture characteristics produced significantly better accuracy, which supported their synergistic significance in the predictive model. The result indicates that the textural characteristics can be an objective, accurate, and comprehensive glioma prediction when paired with conventional imagery.

CONCLUSION

The results outperform current approaches for identifying LGG from HGG and provide competitive performance in classifying four categories of glioma in the literature. The proposed model can help stratify patients in clinical studies, choose patients for targeted therapy, and customize specific treatment schedules.