Purpose
To analyze the implementation of deep learning software for the detection and worklist prioritization of acute intracranial hemorrhage on non-contrast head CT (NCCT) in various clinical ...settings at an academic medical center.
Methods
Urgent NCCT scans were reviewed by the Aidoc (Tel Aviv, Israel) neural network software. All cases flagged by the software as positive for acute intracranial hemorrhage on the neuroradiology worklist were prospectively included in this assessment. The scans were classified regarding presence and type of hemorrhage, whether these were initial or follow-up scans, and patient visit location, including trauma/emergency, inpatient, and outpatient departments.
Results
During the 2 months of enrollment, 373 NCCT scans were flagged by the Aidoc software for possible intracranial hemorrhage out of 2011 scans analyzed (18.5%). Among the flagged cases, 275 (72.4%) were positive; 290 (77.7%) were inpatient cases, 75 (20.1%) were trauma/emergency cases, and eight (2.1%) were outpatient cases, and 229 of 373 (62.5%) were follow-up cases, of which 219 (95.6%) inpatient cases. Among the 144 new cases flagged for hemorrhage, 66 (44.4%) were positive, of which 39 (58.2%) were trauma/emergency cases. The overall sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 88.7%, 94.2% and 73.7%, 97.7%, and 93.4%, respectively. The accuracy of the intracranial hemorrhage detection was significantly higher for emergency cases than for inpatient cases (96.5% versus 89.4%).
Conclusion
This study reveals that the performance of the deep learning software for acute intracranial hemorrhage detection varies depending upon the patient visit location. Furthermore, a substantial portion of flagged cases were follow-up exams, the majority of which were inpatient exams. These findings can help optimize the artificial intelligence-driven clincical workflow.
Advanced imaging analysis for the prediction of tumor biology and modelling of clinically relevant parameters using computed imaging features is part of the emerging field of radiomics research. Here ...we test the hypothesis that a machine learning approach can distinguish grade 1 from higher gradings in meningioma patients using radiomics features derived from a heterogenous multicenter dataset of multi-paramedic MRI.
A total of 138 patients from 5 international centers that underwent MRI prior to surgical resection of intracranial meningiomas were included. Segmentation was performed manually on co-registered multi-parametric MR images using apparent diffusion coefficient (ADC) maps, T1-weighted (T1), post-contrast T1-weighted (T1c), subtraction maps (Sub, T1c – T1), T2-weighted fluid-attenuated inversion recovery (FLAIR) and T2-weighted (T2) images. Feature selection was performed and using cross-validation to separate training from testing data, four machine learning classifiers were scored on combinations of MRI modalities: random forest (RF), extreme gradient boosting (XGBoost), support vector machine (SVM) and multilayer perceptron (MLP).
The best AUC of 0.97 (1.0 and 0.97 for sensitivity and specificity) was observed for the combination of ADC, ADC of the peritumoral edema, T1, T1c, Sub and FLAIR-derived features using only 16 of the 10,914 possible features and XGBoost.
Machine learning using radiomics features derived from multi-parametric MRI is capable of high AUC scores with high sensitivity and specificity in classifying meningiomas between low and higher gradings despite heterogeneous protocols across different centers. Feature selection can be performed effectively even when extracting a large amount of data for radiomics fingerprinting.
MRI plays an important role in the evaluation of glioblastoma, both at initial diagnosis and follow up after treatment. Quantitative analysis
radiomics can augment the interpretation of MRI in terms ...of providing insights regarding the differential diagnosis, genotype, treatment response, and prognosis. The various MRI radiomic features of glioblastoma are reviewed in this article.
Various substances, including methemoglobin, melanin, lipid, protein, calcium, iron, copper, and manganese, are responsible for the intrinsically high signal intensity observed in intracranial ...lesions at T1-weighted magnetic resonance (MR) imaging. Many of these substances have physical properties that lead to other specific imaging features as well. For example, lipid-containing lesions frequently produce chemical shift artifact, and some melanin-containing lesions exhibit a combination of high signal intensity on T1-weighted images and low signal intensity on T2-weighted images. The location and extent of a region of abnormal signal hyperintensity may be helpful for identifying rare diseases such as an ectopic posterior pituitary gland near the floor of the third ventricle, bilateral involvement of the dentate and lentiform nuclei in Cockayne syndrome, and involvement of the anterior temporal lobe and cerebellum in neurocutaneous melanosis. In cases in which diagnostically specific T1-weighted imaging features are lacking, findings obtained with other MR pulse sequences and other modalities can help narrow the differential diagnosis: An elevated glutamine or glutamate level at MR spectroscopy is suggestive of hepatic encephalopathy; a popcorn ball-like appearance at T2-weighted imaging, of cavernous malformations; and hyperattenuation at computed tomography, of mineral deposition disease. In many cases, a comparison of imaging features with clinical measures enables a specific diagnosis.
Cerebellar tonsillar reduction or resection can be performed as part of the surgical management of Chiari type 1 malformation when it is accompanied by symptomatic brainstem compression or ...syringomyelia. The purpose of this study is to characterize the early postoperative magnetic resonance imaging (MRI) findings in patients with Chiari type 1 malformations who undergo cerebellar tonsillar reduction via electrocautery.
The extent of cytotoxic edema and microhemorrhages demonstrated on MRI scans obtained within 9 days following surgery was assessed and correlated with neurological symptoms.
Cytotoxic edema was found on all postoperative MRI examinations included in this series, with superimposed hemorrhage in 12 of 16 patients (75%) and was primarily located along the margins of the cauterized inferior cerebellum. Cytotoxic edema was present beyond the margins of the cauterized cerebellar tonsils in 5 of 16 patients (31%) and was associated with new focal neurological deficits in 4 of 5 patients (80%).
Cytotoxic edema and hemorrhages along the cerebellar tonsil cautery margins can be expected findings in early postoperative MRI in patients who undergo Chiari decompression accompanied by tonsillar reduction. However, the presence of cytotoxic edema beyond these regions can be associated with new focal neurological symptoms.
Imaging Features of Solitary Fibrous Tumors GINAT, Daniel T; BOKHARI, Aqiba; BHATT, Shweta ...
American journal of roentgenology (1976),
03/2011, Letnik:
196, Številka:
3
Journal Article
Recenzirano
The goal of this pictorial essay is to illustrate the multimodality imaging features of pleural and extrapleural solitary fibrous tumors.
Solitary fibrous tumors tend to be well-defined, ovoid, ...heterogeneously enhancing lesions. MRI characteristically depicts areas of low signal intensity that correspond to dense collagen. The findings of lesion multiplicity and hypermetabolism on PET images should raise the suspicion of malignancy.
From a clinical-radiologic standpoint, there are a limited number of structures and disease entities in the temporal bone with which one must be familiar in order to proficiently interpret a computed ...tomographic or magnetic resonance imaging study of the temporal bone. It is helpful to examine the region in an organized and systematic fashion, going through the same checklist of key structures each time. This is the first of a two-part review that provides a practical approach to understanding temporal bone anatomy, localizing a pathologic process with a focus on inflammatory and neoplastic processes, identifying pertinent positives and negatives, and formulating a differential diagnosis.
The objective of our study was to determine the utility of diffusion-weighted imaging (DWI) and cell density for differentiating benign from malignant skull lesions.
A retrospective review was ...performed. Minimum apparent diffusion coefficient (ADC) values were measured and normalized to white matter, which we refer to as "normalized ADC," in 24 skull lesions (12 malignant and 12 benign) in 18 patients. In addition, cell densities were measured in 15 cases and correlated with ADC values.
The average minimum ADC in malignant tumors was 0.70 × 10(-3) mm(2)/s versus 1.11 × 10(-3) mm(2)/s in benign tumors (p = 0.0037). Similarly, the average normalized ADC for malignant tumors was 1.03, whereas the average normalized ADC for benign tumors was 1.65 (p = 0.0012). Receiver operating characteristic curve analysis yielded optimal normalized ADC and ADC thresholds of 1.23 (accuracy, 84.6%; sensitivity, 75.0%; specificity, 92.3%) and 1.01 × 10(-3) mm(2)/s (accuracy, 83.7%; sensitivity, 83.3%; specificity, 84.6%), respectively. There was a significant inverse correlation between cell density and normalized ADC (r = -0.58; p = 0.023). The low cellularity in chordoma and low-grade chondrosarcoma and high cellularity in eosinophilic granuloma may explain the DWI features of these lesions.
ADC values in skull lesions correlate with cell density and can potentially narrow the differential diagnoses for indeterminate skull lesions. Understanding the histopathologic features of skull lesions can refine interpretation of DWI.
The first part of this review of the temporal bone discussed anatomy of the temporal bone as well as inflammatory and neoplastic processes in the temporal bone region (1). This second part will first ...discuss trauma to the temporal bone and posttraumatic complications. The indications for common surgical procedures performed in the temporal bone and their postoperative imaging appearance are then presented. Finally, a few noninflammatory nonneoplastic entities involving the temporal bone are reviewed. They are relatively uncommon diagnoses compared with infectious or inflammatory diseases. However, because patients present with symptoms that are either common (hearing loss) or distinctive (sensorineural hearing loss in a child), they are important for the radiologist to be aware of and recognize.