Toward Automated Detection of Silent Cerebral Infarcts in Children and Young Adults With Sickle Cell Anemia.

BACKGROUND: Silent cerebral infarcts (SCI) in sickle cell anemia (SCA) are associated with future strokes and cognitive impairment, warranting early diagnosis and treatment. Detection of SCI, however, is limited by their small size, especially when neuroradiologists are unavailable. We hypothesized that deep learning may permit automated SCI detection in children and young adults with SCA as a tool to identify the presence and extent of SCI in clinical and research settings. METHODS: We utilized UNet-a deep learning model-for fully automated SCI segmentation. We trained and optimized UNet using brain magnetic resonance imaging from the SIT trial (Silent Infarct Transfusion). Neuroradiologists provided the ground truth for SCI diagnosis, while a vascular neurologist manually delineated SCI on fluid-attenuated inversion recovery and provided the ground truth for SCI segmentation. UNet was optimized for the highest spatial overlap between automatic and manual delineation (dice similarity coefficient). The optimized UNet was externally validated using an independent single-center prospective cohort of SCA participants. Model performance was evaluated through sensitivity and accuracy (%correct cases) for SCI diagnosis, dice similarity coefficient, intraclass correlation coefficient (metric of volumetric agreement), and Spearman correlation. RESULTS: The SIT trial (n=926; 31% with SCI; median age, 8.9 years) and external validation (n=80; 50% with SCI; age, 11.5 years) cohorts had small median lesion volumes of 0.40 and 0.25 mL, respectively. Compared with the neuroradiology diagnosis, UNet predicted SCI presence with 100% sensitivity and 74% accuracy. In magnetic resonance imaging with SCI, UNet reached a moderate spatial agreement (dice similarity coefficient, 0.48) and high volumetric agreement (intraclass correlation coefficient, 0.76; ρ=0.72; P<0.001) between automatic and manual segmentations. CONCLUSIONS: UNet, trained using a large pediatric SCA magnetic resonance imaging data set, sensitively detected small SCI in children and young adults with SCA. While additional training is needed, UNet may be integrated into the clinical workflow as a screening tool, aiding in SCI diagnosis.

Evaluating the effect of a post-processing algorithm in detection of annular fissure on MR imaging.

BACKGROUND AND PURPOSE: Visualization of annular fissures on MRI is becoming increasingly important but remains challenging. Our purpose was to test whether an image processing algorithm could improve detection of annular fissures. MATERIALS AND METHODS: In this retrospective study, two neuroradiologists identified 56 IVDs with annular fissures and 97 IVDs with normal annulus fibrosus in lumbar spine MRIs of 101 patients (58 M, 43 F; age ± SD 15.1 ± 3.0 years). Signal intensities of diseased and normal annulus fibrosus, and contrast-to-noise ratio between them on sagittal T2-weighted images were calculated before and after processing with a proprietary software. Effect of processing on detection of annular fissures by two masked neuroradiologists was also studied for IVDs with Pfirrmann grades of ≤ 2 and > 2. RESULTS: Mean (SD) signal baseline intensities of diseased and normal annulus fibrosus were 57.6 (23.3) and 24.4 (7.8), respectively (p < 0.001). Processing increased (p < 0.001) the mean (SD) intensity of diseased annulus to 110.6 (47.9), without affecting the signal intensity of normal annulus (p = 0.14). Mean (SD) CNR between the diseased and normal annulus increased (p < 0.001) from 11.8 (14.1) to 29.6 (29.1). Both masked readers detected more annular fissures after processing in IVDs with Pfirrmann grade of ≤ 2 and > 2, with an apparent increased sensitivity and decreased specificity using predefined image-based human categorization as a reference standard. CONCLUSIONS: Image processing improved CNR of annular fissures and detection rate of annular fissures. However, further studies with a more stringent reference standard are needed to assess its effect on sensitivity and specificity.

ACR Appropriateness Criteria Head Trauma.

Neuroimaging plays an important role in the management of head trauma. Several guidelines have been published for identifying which patients can avoid neuroimaging. Noncontrast head CT is the most appropriate initial examination in patients with minor or mild acute closed head injury who require neuroimaging as well as patients with moderate to severe acute closed head injury. In short-term follow-up neuroimaging of acute traumatic brain injury, CT and MRI may have complementary roles. In subacute to chronic traumatic brain injury, MRI is the most appropriate initial examination, though CT may have a complementary role in select circumstances. Advanced neuroimaging techniques are areas of active research but are not considered routine clinical practice at this time. In suspected intracranial vascular injury, CT angiography or venography or MR angiography or venography is the most appropriate imaging study. In suspected posttraumatic cerebrospinal fluid leak, high-resolution noncontrast skull base CT is the most appropriate initial imaging study to identify the source, with cisternography reserved for problem solving. The ACR Appropriateness Criteria are evidence-based guidelines for specific clinical conditions that are reviewed every three years by a multidisciplinary expert panel. The guideline development and review include an extensive analysis of current medical literature from peer-reviewed journals and the application of a well-established consensus methodology (modified Delphi) to rate the appropriateness of imaging and treatment procedures by the panel. In those instances in which evidence is lacking or not definitive, expert opinion may be used to recommend imaging or treatment.