Texture Analysis of Brain CT Scans for ICP Prediction

Abstract

Elevated Intracranial Pressure (ICP) is a significant cause of mortality and long-term functional damage in traumatic brain injury (TBI). Current ICP monitoring methods are highly invasive, presenting additional risks to the patient. This paper describes a computerized noninvasive screening method based on texture analysis of computed tomography (CT) scans of the brain, which may assist physicians in deciding whether to begin invasive monitoring. Quantitative texture features extracted using statistical, histogram and wavelet transform methods are used to characterize brain tissue windows in individual slices, and aggregated across the scan. Support Vector Machine (SVM) is then used to predict high or normal levels of ICP using the most significant features from the aggregated set. Results are promising, providing over 80% predictive accuracy and good separation of the two ICP classes, confirming the suitability of the approach and the predictive power of texture features in screening patients for high ICP.

AI-Based Decision Support System for Traumatic Brain Injury: A Survey

“…Textural-based methods assess subtle changes in tissue density and image texture that are sometimes hidden from human eyes to automate ICP prediction. A combination of features extracted from Fourier analysis, Gray Level Run Length Matrix (GLRLM), Dual Tree Complex Wavelet Transform (DT-CWT), and histogram analysis have been used for ICP level classification in TBI patients [77][78][79][80][81]. It was demonstrated that the energy of different sub-band images of 2D fully anisotropic Morlet wavelet transformations could be used to determine the dominant textural orientation of the brain tissue in TBI patients and was later shown to be more competent than DT-CWT in ICP prediction [82,83].…”

Automated Detection and Screening of Traumatic Brain Injury (TBI) Using Computed Tomography Images: A Comprehensive Review and Future Perspectives

“…Chen et al [ 29 ] presented a texture-based approach to categorize the ICP levels of the CT scans into high versus normal using a threshold value of 15 mmHg. The proposed study applied a machine learning technique that extracts a set of 10 optimized features, and obtained a classification accuracy of 80% using SVM.…”

“…Moreover, manual segmentation of hematoma or midline structures from selected CT slices is challenging due to reasons such as variation in pixel-wise intensity, uneven boundaries, high contrast of tissues, and the presence of noise and artefact [ 26 , 27 , 28 ]. Furthermore, the set of features required for CT-based ICP estimation cannot be readily identified by visual inspection, and is also subject to intra-observer and inter-observer variability [ 29 ]. Moreover, the measurement of MLS should be carried out at the level of the foramen of Monro based on clinical guidelines, and hence, the selection of the appropriate CT slice is crucial [ 30 ].…”

Brain tumors detection and segmentation in MR images: Gabor wavelet vs. statistical features

“…Entropy reveals the randomness of intensity values. Formulae for these features are listed as follows [22]:…”