Purpose: To validate Bridge Burner, a new brain segmentation algorithm based on thresholding, connectivity, surface detection, and a new operator of constrained growing. Materials and Methods:T1-weighted MR images were selected at random from three previous neuroimaging studies to represent a spectrum of system manufacturers, pulse sequences, subject ages, genders, and neurological conditions. The ground truth consisted of brain masks generated manually by a consensus of expert observers. All cases were segmented using a common set of parameters.Results: Bridge Burner segmentation errors were 3.4% Ϯ 1.3% (volume mismatch) and 0.34 Ϯ 0.17 mm (surface mismatch). The disagreement among experts was 3.8% Ϯ 2.0% (volume mismatch) and 0.48 Ϯ 0.49 mm (surface mismatch). The error obtained using the brain extraction tool (BET), a widely used brain segmentation program, was 8.3% Ϯ 9.1%. Bridge Burner brain masks are visually similar to the masks generated by human experts. Areas affected by signal intensity nonuniformity artifacts were occasionally undersegmented, and meninges and large sinuses were often falsely classified as the brain tissue. Segmentation of one MRI dataset takes seven seconds. Conclusion:The new fully automatic algorithm appears to provide accurate brain segmentation from high-resolution T1-weighted MR images. SEGMENTATION OF THE BRAIN is an important preprocessing step in neuroimaging applications. Total brain volumes and subsequent brain atrophy estimation in patients suffering from various pathologies, including traumatic injury, multiple sclerosis, or dementia, are useful estimates of brain injury and its response to treatment (1,2). Brain segmentation is the initial step in studies of the global and regional brain shape and volume (3-7). These studies are taking place with increasing frequency in research on normal brain development during childhood, normal aging, and neurological and psychiatric disorders. Coregistration of functional brain data with high-resolution MR or computed tomography (CT) images is another area that benefits from brain segmentation (8,9). There is little nonbrain tissue signal in functional imaging such as positron emission tomography (PET), single-photon emission CT (SPECT), or functional MRI (fMRI), whereas structural MR images may contain high signal intensity from nonbrain tissue. As a result, most multimodality registration algorithms work best when processing of structural data is restricted to brain voxels. Segmentation of the brain is also a key step in cortical surface modeling and visualization.The tedious and expensive nature of manual outlining of the brain provided the impetus for the development of several automated and semiautomated skullstripping systems. Several sophisticated algorithms have been developed and extensively tested in recent publications (10 -14). These programs have greatly reduced the amount of time needed to segment brain tissue compared to manual extraction. However, the majority of algorithms include ventricular and subarachnoid cerebro...
Genotype, particularly Ras status, greatly affects prognosis and treatment of liver metastasis in colon cancer patients. This pilot aimed to apply word frequency analysis and a naive Bayes classifier on radiology reports to extract distinguishing imaging descriptors of wild-type colon cancer patients and those with v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS) mutations. In this institutional-review-board-approved study, we compiled a SNaPshot mutation analysis dataset from 457 colon adenocarcinoma patients. From this cohort of patients, we analyzed radiology reports of 299 patients (> 32,000 reports) who either were wild-type (147 patients) or had a KRAS (152 patients) mutation. Our algorithm determined word frequency within the wild-type and mutant radiology reports and used a naive Bayes classifier to determine the probability of a given word belonging to either group. The classifier determined that words with a greater than 50% chance of being in the KRAS mutation group and which had the highest absolute probability difference compared to the wild-type group included: “several”, “innumerable”, “confluent”, and “numerous” (p < 0.01). In contrast, words with a greater than 50% chance of being in the wild type group and with the highest absolute probability difference included: “few”, “discrete”, and “[no] recurrent” (p = 0.03). Words used in radiology reports, which have direct implications on disease course, tumor burden, and therapy, appear with differing frequency in patients with KRAS mutations versus wild-type colon adenocarcinoma. Moreover, likely characteristic imaging traits of mutant tumors make probabilistic word analysis useful in identifying unique characteristics and disease course, with applications ranging from radiology and pathology reports to clinical notes.
No abstract
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.