Noncontrast Head Computed Tomography Research Articles

Variations of Tentorial Notch Anatomy in Autopsy and NCCT of Head Injury Patients to Correlate Its Impact over Brainstem—An Observational Study

Abstract Background Tentorium cerebelli divide cranial cavity into supratentorial and infratentorial compartment. Tentorial notch is a gap in tentorium cerebelli located centrally and anteriorly, related to upper brainstem, first six cranial nerves, cerebrum, and cerebellum. The aim of this study was to analyze morphometric variation in tentorial notch anatomy in autopsy and computed tomography (CT) of head injury patients and to find out correlation between tentorial notch anatomy and uncal herniation and changes over brainstem surface in case of uncal herniation. Materials and Methods Autopsy examination of head injury patients was done between July 1, 2021, and Jan 31, 2023. Skull was opened in a standard manner and midbrain sectioned at the level of tentorial edge. Tentorial notch parameters (anterior notch width, maximum notch width, notch length, posterior tentorial length, apicotectal distance, interpedunculoclival distance) were measured using geometry compass and vernier caliper. These parameters were also measured on noncontrast computed tomography (NCCT) head of same patients. Evidence of uncal herniation was sought for in both NCCT head and autopsy examination. Impact of uncal herniation over brainstem in form of indentation over lateral surface of midbrain and macroscopic brainstem hemorrhage was noted. Results In 65 autopsy cases, there were 56 male and 9 female specimens; their mean age was 40.63 ± 16.78 years (range: 7–86 years). The most frequent type of tentorial notch observed in our study was typical type (32.30%). Relative frequency of uncal herniation was present predominantly (75%) in large type of tentorial notch and least (20%) in small type of tentorial notch. A significant correlation (p-value < 0.01) was found between uncal herniation and indentation over lateral surface of midbrain, while no significant correlation (p-value > 0.05) was found between uncal herniation and macroscopic brainstem hemorrhage in autopsy specimens. Conclusion The results of our study provide a baseline data about tentorial notch anatomy and it may facilitate neurosurgical decision making as well as help in deciding the best trajectory for lesions approaching in the vicinity of the tentorial notch

Assessment of image quality and impact of deep learning-based software in non-contrast head CT scans

In this retrospective study, we aimed to assess the objective and subjective image quality of different reconstruction techniques and a deep learning-based software on non-contrast head computed tomography (CT) images. In total, 152 adult head CT scans (77 female, 75 male; mean age 69.4 ± 18.3 years) obtained from three different CT scanners using different protocols between March and April 2021 were included. CT images were reconstructed using filtered-back projection (FBP), iterative reconstruction (IR), and post-processed using a deep learning-based algorithm (PS). Post-processing significantly reduced noise in FBP-reconstructed images (up to 15.4% reduction) depending on the protocol, leading to improvements in signal-to-noise ratio of up to 19.7%. However, when deep learning-based post-processing was applied to FBP images compared to IR alone, the differences were inconsistent and partly non-significant, which appeared to be protocol or site specific. Subjective assessments showed no significant overall improvement in image quality for all reconstructions and post-processing. Inter-rater reliability was low and preferences varied. Deep learning-based denoising software improved objective image quality compared to FBP in routine head CT. A significant difference compared to IR was observed for only one protocol. Subjective assessments did not indicate a significant clinical impact in terms of improved subjective image quality, likely due to the low noise levels in full-dose images.

Estimate and compensate head motion in non-contrast head CT scans using partial angle reconstruction and deep learning.

Patient head motion is a common source of image artifacts in computed tomography (CT) of the head, leading to degraded image quality and potentially incorrect diagnoses. The partial angle reconstruction (PAR) means dividing the CT projection into several consecutive angular segments and reconstructing each segment individually. Although motion estimation and compensation using PAR has been developed and investigated in cardiac CT scans, its potential for reducing motion artifacts in head CT scans remains unexplored. To develop a deep learning (DL) model capable of directly estimating head motion from PAR images of head CT scans and to integrate the estimated motion into an iterative reconstruction process to compensate for the motion. Head motion is considered as a rigid transformation described by six time-variant variables, including the three variables for translation and three variables for rotation. Each motion variable is modeled using a B-spline defined by five control points (CP) along time. We split the full projections from 360° into 25 consecutive PARs and subsequently input them into a convolutional neural network (CNN) that outputs the estimated CPs for each motion variable. The estimated CPs are used to calculate the object motion in each projection, which are incorporated into the forward and backprojection of an iterative reconstruction algorithm to reconstruct the motion-compensated image. The performance of our DL model is evaluated through both simulation and phantom studies. The DL model achieved high accuracy in estimating head motion, as demonstrated in both the simulation study (mean absolute error (MAE) ranging from 0.28 to 0.45mm or degree across different motion variables) and the phantom study (MAE ranging from 0.40 to 0.48mm or degree). The resulting motion-corrected image, , exhibited a significant reduction in motion artifacts when compared to the traditional filtered back-projection reconstructions, which is evidenced both in the simulation study (image MAE drops from 178 33HU to 37 9HU, structural similarity index (SSIM) increases from 0.60 0.06 to 0.98 0.01) and the phantom study (image MAE drops from 117 17HU to 42 19HU, SSIM increases from 0.83 0.04 to 0.98 0.02). We demonstrate that using PAR and our proposed deep learning model enables accurate estimation of patient head motion and effectively reduces motion artifacts in the resulting head CT images.

The prevalence of non-contrast CT imaging abnormalities in reversible cerebral vasoconstriction syndrome: A systematic review and meta-analysis.

Reversible cerebral vasoconstriction syndrome (RCVS) is a syndrome of recurrent thunderclap headaches and reversible vasoconstriction of the cerebral arteries on neuroimaging within 3 months of onset. Initial non-contrast computed tomography (CT) can reveal abnormalities such as ischemic stroke, intracerebral hemorrhage, and subarachnoid hemorrhage (SAH) can be present in patients with RCVS and may delay diagnosis. We conducted a systematic review and meta-analysis in accordance with the PRISMA guidelines. We aimed to estimate the prevalence of imaging abnormalities on initial non-contrast CT head in adult patients with RCVS. We searched electronic databases including MEDLINE, EMBASE, and the Cochrane Register of Clinical Trials from inception to August 2, 2022. Eligible studies included articles reporting the prevalence of non-contrast CT abnormalities on initial neuroimaging in patients with RCVS, aged 18 and older. Case series, observational studies and clinical trials were included. Data was extracted directly from included papers using a standardized data charting form. The search yielded 722 titles with duplicates removed. Twenty studies that included 379 patients with RCVS met inclusion criteria. We classified non-contrast CT abnormalities as either ischemic stroke, ICH, or SAH. We pooled prevalence data using a random effects model with the inverse-variance weighted method. The most common imaging finding was SAH with a pooled prevalence of 24% (95% CI:17%-33%), followed by ICH at 14% (95% CI:8%-22%), and ischemic stroke at 10% (95% CI:7%-14%). The pooled prevalence of any of these imaging abnormalities on initial non-contrast CT was 31% (95% CI:23%-40%). Risk of bias was moderate to very-high-risk for case-series and low-risk for observational studies. Our review demonstrates that one-third of patients with RCVS will have an abnormality on initial non-contrast CT head, including either an ischemic stroke, ICH, or SAH. These findings highlight the diagnostic challenges of RCVS imaging and contribute to our understanding of this disease.

Craniofacial Injury in a Child by Agricultural Fan Blade: A Case Report

AbstractFarm machine-related injuries in children are common modality of head injury. We describe here a case of threshing fan blade injury to head and its successful management. A 10-year-old male was brought to the emergency department of All India Institute of Medical Sciences (AIIMS) Patna with the complains of accidental fall on the fan blade of the thresher machine. On arrival he had a Glasgow Coma Scale of E1VTM5 status, bilateral pupil mid-dilated and sluggish reaction to light. On secondary survey, he had a lacerated wound of size 15 × 4 × 2 cm extending from left side forehead to right temporal region exposing bone and brain parenchyma herniation through the wound. A noncontrast computed tomography head showed B/L frontal contusion with gross edema, subarachnoid hemorrhage, and intraventricular hemorrhage. There was fracture of bilateral frontal bone extending to right parietotemporal bone. Patient was taken to operating room, where after painting and draping, the wound was irrigated with normal saline. The herniated brain was removed, securing hemostasis with surgical and Bovie cautery. He underwent a prolonged neurorehabilitation and made a complete recovery after 3 months of injury. Bihar is an eastern part of India where agriculture plays a major role in rural livelihood. Winnowing fan machines are used to separate the grain from the husk. Children have peculiar behavior to inspect things and this curious behavior leads to coming in contacts with the open blade. Falling on even a stationary blade can sometimes lead to head injury. Early surgery with meticulous wound debridement and closure of dura is required. Covering the fan blade as well as careful parenting of children will prevent further injuries.

Uncertainty-aware deep-learning model for prediction of supratentorial hematoma expansion from admission non-contrast head computed tomography scan

Hematoma expansion (HE) is a modifiable risk factor and a potential treatment target in patients with intracerebral hemorrhage (ICH). We aimed to train and validate deep-learning models for high-confidence prediction of supratentorial ICH expansion, based on admission non-contrast head Computed Tomography (CT). Applying Monte Carlo dropout and entropy of deep-learning model predictions, we estimated the model uncertainty and identified patients at high risk of HE with high confidence. Using the receiver operating characteristics area under the curve (AUC), we compared the deep-learning model prediction performance with multivariable models based on visual markers of HE determined by expert reviewers. We randomly split a multicentric dataset of patients (4-to-1) into training/cross-validation (n = 634) versus test (n = 159) cohorts. We trained and tested separate models for prediction of ≥6 mL and ≥3 mL ICH expansion. The deep-learning models achieved an AUC = 0.81 for high-confidence prediction of HE≥6 mL and AUC = 0.80 for prediction of HE≥3 mL, which were higher than visual maker models AUC = 0.69 for HE≥6 mL (p = 0.036) and AUC = 0.68 for HE≥3 mL (p = 0.043). Our results show that fully automated deep-learning models can identify patients at risk of supratentorial ICH expansion based on admission non-contrast head CT, with high confidence, and more accurately than benchmark visual markers.

Contribution of Deep Cerebral Venous Anomaly to the Emergence of Nonaneurysmal Subarachnoid Hemorrhage as Opposed to Aneurysmal Subarachnoid Hemorrhage

The exact reason of nonaneurysmal subarachnoid hemorrhage (SAH) is an enigma. The aim of this study is to identify if type III deep cerebral venous drainage is exclusively prevalent in patients with nonaneurysmal SAH and to enumerate the predictors of poorer outcome in these patients. All patients of age >18years, presented at our centre with spontaneous SAH on noncontrast computed tomography head and were divided into 2 groups, aneurysmal and nonaneurysmal SAH after 4-vessel DSA. Based on the deep venous drainage pattern on both sides, basal venous drainage was found and classified into 3 types: type I, type II, and type III. The 3 groups were pitted against one another. Regression analysis were performed to predict the occurrence of nonaneurysmal-SAH with different types of basal vein. There were 100 nonaneurysmal SAH cases and 103 aneurysmal SAH cases. The mean age of presentation was 47.8±13.55years with slight male predominance (52%). The patients with type III venous drainage have 2 times more risk of developing nonaneurysmal SAH (95% confidence interval=1.21-4.31) as compared to those with aneurysmal SAH. On multivariate analysis, type III basal venous drainage, worse Hunt and Hess grade at presentation, extensive bleeding were predictors of an adverse outcome. The presence of type III venous distribution is associated with a 2-fold increase in the probability of having nonaneurysmal SAH, as well as a 3-fold increase in the risk of developing poorer neurological sequelae.

Decreasing Lens Irradiation on Brain Imaging: A Multi-CT Scanner Quality Improvement Project.

Cataracts, a leading global cause of blindness, are associated with ionising radiation exposure. This audit aimed to enhance lens exclusion during non-contrast head computed tomography (CT) scans at Newham University Hospital (NUH) using two CT scanners. A retrospective audit of non-contrast head CT scans at NUH excluded scans for trauma and imaging of orbital structures. A one-week audit in April 2023 assessed lens exclusion, compared to the Royal College of Radiologists (RCR) standards. A total of 101 consecutive scans were analysed and 63 (62%) scans were included in the final study. Thirty-eight percent of the scans were excluded according to the exclusion criteria of head, neckand facial traumas, orbital infectionsand papilledema. Results were presented, followed by a three-month radiographer re-education period, emphasizing gantry tilt and patient positioning. A reaudit in August 2023 evaluated outcomes. For the reaudit, 183 consecutive scans were analysed, with 131 (72%) scans included in the final studyand 52 (28%) scans excluded according to the same exclusion criteria as the first audit. Lens exclusion in non-contrast head CT scans improved significantly from 0/63 (0%) compliance to 19/131 (14.50%) (p=0005) compliance with the standards. Variability in radiographer practices, 'near misses' and time constraints were identified as challenges. Staff turnover impacted compliance. This audit has shed light on a critical aspect of patient care in the field of radiology. This research underscores the importance of rigorous and standardised protocols in radiological procedures, particularly when it comes to protecting the lens of the eye. By enhancing lens exclusion during non-contrast head CT scans, we have taken a significant step in mitigating the risk associated with ionising radiation exposure. Although substantial improvements were made, achieving the RCR audit standard remained elusive. Ongoing re-education, reauditsand a multidisciplinary approach are necessary to optimise radiographer adherence and reduce ionising radiation exposure to the lens during head CT scans.This quality improvement project proves thatcontinued emphasis on gantry tilt and patient positioning in radiographer education and training can make a significant difference in patient safety.As we move forward, let us remember that even small improvements can make a big difference in safeguarding the health and well-being of patients.

Diagnostic efficiency of individual systems for automatic analysis of computed tomography images in the detection of ischemic stroke in the basin of the middle cerebral artery

Background. The diagnosis of ischemic stroke is of high importance in modern medical practice. One of the most promising methods for solving this problem is the introduction of machine learning algorithms into physicians’ work as an auxiliary tool for the interpretation of beam images.Aim: To compare automated computed tomography (CT) image analysis systems in detecting middle cerebral artery stroke.Material and Methods. The study included three anonymized (A, B, C) machine learning algorithms. Analytical validation was carried out on a database of one hundred patients admitted in St. Petersburg vascular center with suspected middle cerebral artery stroke, who underwent noncontrast head CTs. Ischemic stroke in half of the patients was confirmed on the basis of clinical examination findings and CT-angiography and CT-perfusion. The study evaluated the performance indicators (sensitivity, specificity, positive predictive value, negative predictive value, accuracy) for detecting a set of signs of early ischemic changes (by automatic segmentation and predicting a score on the ASPECTS scale). The article also provides a graph that allows you to evaluate the quality of a binary classification – characteristic curves (ROC-curves).Results. The meta-analyses showed all the considered automated algorithms did not reach the threshold values of accuracy (range from 0.67 to 0.75) required for programs according to clinical guidelines (0.80). The algorithms showed variability in sensitivity and specificity. One of the automatic analysis systems (A) had a high sensitivity (0.88), but at the same time a low specificity (0.46), which indicates its overtraining and a tendency to overdiagnoses. The remaining algorithms (B, C) showed low sensitivity (0.6; 0.55) and high specificity (0.9; 0.8).

Radiomic markers of intracerebral hemorrhage expansion on non-contrast CT: independent validation and comparison with visual markers.

To devise and validate radiomic signatures of impending hematoma expansion (HE) based on admission non-contrast head computed tomography (CT) of patients with intracerebral hemorrhage (ICH). Utilizing a large multicentric clinical trial dataset of hypertensive patients with spontaneous supratentorial ICH, we developed signatures predictive of HE in a discovery cohort (n = 449) and confirmed their performance in an independent validation cohort (n = 448). In addition to n = 1,130 radiomic features, n = 6 clinical variables associated with HE, n = 8 previously defined visual markers of HE, the BAT score, and combinations thereof served as candidate variable sets for signatures. The area under the receiver operating characteristic curve (AUC) quantified signatures' performance. A signature combining select radiomic features and clinical variables attained the highest AUC (95% confidence interval) of 0.67 (0.61-0.72) and 0.64 (0.59-0.70) in the discovery and independent validation cohort, respectively, significantly outperforming the clinical (pdiscovery = 0.02, pvalidation = 0.01) and visual signature (pdiscovery = 0.03, pvalidation = 0.01) as well as the BAT score (pdiscovery < 0.001, pvalidation < 0.001). Adding visual markers to radiomic features failed to improve prediction performance. All signatures were significantly (p < 0.001) correlated with functional outcome at 3-months, underlining their prognostic relevance. Radiomic features of ICH on admission non-contrast head CT can predict impending HE with stable generalizability; and combining radiomic with clinical predictors yielded the highest predictive value. By enabling selective anti-expansion treatment of patients at elevated risk of HE in future clinical trials, the proposed markers may increase therapeutic efficacy, and ultimately improve outcomes.

Real world validation of an AI-based CT hemorrhage detection tool.

Intracranial hemorrhage (ICH) is a potentially life-threatening medical event that requires expedited diagnosis with computed tomography (CT). Automated medical imaging triaging tools can rapidly bring scans containing critical abnormalities, such as ICH, to the attention of radiologists and clinicians. Here, we retrospectively investigated the real-world performance of VeriScout™, an artificial intelligence-based CT hemorrhage detection and triage tool. Ground truth for the presence or absence of ICH was iteratively determined by expert consensus in an unselected dataset of 527 consecutively acquired non-contrast head CT scans, which were sub-grouped according to the presence of artefact, post-operative features and referral source. The performance of VeriScout™ was compared with the ground truths for all groups. VeriScout™ detected hemorrhage with a sensitivity of 0.92 (CI 0.84-0.96) and a specificity of 0.96 (CI 0.94-0.98) in the global dataset, exceeding the sensitivity of general radiologists (0.88) with only a minor relative decrement in specificity (0.98). Crucially, the AI tool detected 13/14 cases of subarachnoid hemorrhage, a potentially fatal condition that is often missed in emergency department settings. There was no decrement in the performance of VeriScout™ in scans containing artefact or postoperative change. Using an integrated informatics platform, VeriScout™ was deployed into the existing radiology workflow. Detected hemorrhage cases were flagged in the hospital radiology information system (RIS) and relevant, annotated, preview images made available in the picture archiving and communications system (PACS) within 10 min. AI-based radiology worklist prioritization for critical abnormalities, such as ICH, may enhance patient care without adding to radiologist or clinician burden.

Time ratio disparities among ED patients undergoing head CT.

To assess if patients who underwent head computed tomography (CT) experienced disparities in the emergency department (ED) and if the indication for head CT affected disparities. This study employed a retrospective, IRB-approved cohort design encompassing four hospitals. All ED patients between January 2016 and September 2020 who underwent non-contrast head CTs were included. Furthermore, key time intervals including ED length of stay (LOS), ED assessment time, image acquisition time, and image interpretation time were calculated. Time ratio (TR) was used to compare these time intervals between the groups. A total of 45,177 ED visits comprising 4730 trauma cases, 5475 altered mental status cases, 11,925 cases with head pain, and 23,047 cases with other indications were included. Females had significantly longer ED LOS, ED assessment time, and image acquisition time (TR = 1.012, 1.051, 1.018, respectively, P-value < 0.05). This disparity was more pronounced in female patients with head pain complaints compared to their male counterparts (TR = 1.036, 1.059, and 1.047, respectively, P-value < 0.05). Black patients experienced significantly longer ED LOS, image acquisition time, and image assessment time (TR = 1.226, 1.349, and 1.190, respectively, P-value < 0.05). These disparities persisted regardless of head CT indications. Furthermore, patients with Medicare/Medicaid insurance also faced longer wait times in all the time intervals (TR > 1, P-value < 0.001). Wait times for ED head CT completion were longer for Black patients and Medicaid/Medicare insurance holders. Additionally, females experienced extended wait times, particularly when presented with head pain complaints. Our findings underscore the importance of exploring and addressing the contributing factors to ensure equitable and timely access to imaging services in the ED.

Validation of a deep learning model for traumatic brain injury detection and NIRIS grading on non-contrast CT: a multi-reader study with promising results and opportunities for improvement.

This study aimed to assess and externally validate the performance of a deep learning (DL) model for the interpretation of non-contrast computed tomography (NCCT) scans of patients with suspicion of traumatic brain injury (TBI). This retrospective and multi-reader study included patients with TBI suspicion who were transported to the emergency department and underwent NCCT scans. Eight reviewers, with varying levels of training and experience (two neuroradiology attendings, two neuroradiology fellows, two neuroradiology residents, one neurosurgery attending, and one neurosurgery resident), independently evaluated NCCT head scans. The same scans were evaluated using the version 5.0 of the DL model icobrain tbi. The establishment of the ground truth involved a thorough assessment of all accessible clinical and laboratory data, as well as follow-up imaging studies, including NCCT and magnetic resonance imaging, as a consensus amongst the study reviewers. The outcomes of interest included neuroimaging radiological interpretation system (NIRIS) scores, the presence of midline shift, mass effect, hemorrhagic lesions, hydrocephalus, and severe hydrocephalus, as well as measurements of midline shift and volumes of hemorrhagic lesions. Comparisons using weighted Cohen's kappa coefficient were made. The McNemar test was used to compare the diagnostic performance. Bland-Altman plots were used to compare measurements. One hundred patients were included, with the DL model successfully categorizing 77 scans. The median age for the total group was 48, with the omitted group having a median age of 44.5 and the included group having a median age of 48. The DL model demonstrated moderate agreement with the ground truth, trainees, and attendings. With the DL model's assistance, trainees' agreement with the ground truth improved. The DL model showed high specificity (0.88) and positive predictive value (0.96) in classifying NIRIS scores as 0-2 or 3-4. Trainees and attendings had the highest accuracy (0.95). The DL model's performance in classifying various TBI CT imaging common data elements was comparable to that of trainees and attendings. The average difference for the DL model in quantifying the volume of hemorrhagic lesions was 6.0mL with a wide 95% confidence interval (CI) of - 68.32 to 80.22, and for midline shift, the average difference was 1.4mm with a 95% CI of - 3.4 to 6.2. While the DL model outperformed trainees in some aspects, attendings' assessments remained superior in most instances. Using the DL model as an assistive tool benefited trainees, improving their NIRIS score agreement with the ground truth. Although the DL model showed high potential in classifying some TBI CT imaging common data elements, further refinement and optimization are necessary to enhance its clinical utility.

Intraoperative Central Diabetes Insipidus during Aneurysmal Clipping Surgery: An Unusual Phenomenon

AbstractCentral diabetes insipidus (DI) is a known complication associated with pituitary surgeries occurring in postoperative period. However, development of DI following aneurysmal subarachnoid hemorrhage (SAH) is rarely reported. We describe here a case of intraoperative DI in a patient undergoing aneurysmal clipping surgery that posed a challenge for both diagnosis and management. Intraoperative development of central DI was attributed to the evolving ischemic injury to the hypothalamus at the time of rebleeding that was not apparent in the preoperative scan. DI resolved postoperatively after 18 hours of medical management. A careful observation of preoperative scans and vigilant monitoring may help in early diagnosis and management of such complication in the perioperative period.

Artificial intelligence for detection of intracranial haemorrhage on head computed tomography scans: diagnostic accuracy in Hong Kong.

The use of artificial intelligence (AI) to identify acute intracranial haemorrhage (ICH) on computed tomography (CT) scans may facilitate initial imaging interpretation in the accident and emergency department. However, AI model construction requires a large amount of annotated data for training, and validation with real-world data has been limited. We developed an algorithm using an open-access dataset of CT slices, then assessed its utility in clinical practice by validating its performance on CT scans from our institution. Using a publicly available international dataset of >750 000 expert-labelled CT slices, we developed an AI model which determines ICH probability for each CT scan and nominates five potential ICH-positive CT slices for review. We validated the model using retrospective data from 1372 non-contrast head CT scans (84 [6.1%] with ICH) collected at our institution. The model achieved an area under the curve of 0.842 (95% confidence interval=0.791-0.894; P<0.001) for scan-based detection of ICH. A pre-specified probability threshold of ≥50% for the presence of ICH yielded 78.6% accuracy, 73% sensitivity, 79% specificity, 18.6% positive predictive value, and 97.8% negative predictive value. There were 62 true-positive scans and 22 false-negative scans, which could be reduced to six false-negative scans by manual review of model-nominated CT slices. Our model exhibited good accuracy in the CT scan-based detection of ICH, considering the low prevalence of ICH in Hong Kong. Model refinement to allow direct localisation of ICH will facilitate the use of AI solutions in clinical practice.

Headache, vision changes and cranial nerve palsy after head trauma.

A 43-year-old female was brought to the hospital after a ground level fall and her Glasgow Coma Scale was 15. Non-contrast head computed tomography (CT) scan showed small bilateral subdural hematomas, left traumatic subarachnoid hemorrhage, left occipital bone and left longitudinal temporal bone fractures. She was managed conservatively. Interval repeat non-contrast head CT scans showed resolution of the hemorrhage. Two and a half months later she presented to the hospital with headaches, intermittent diplopia, blurry vision and a right lateral gaze palsy. Non-contrast head CT scan showed mild hyperattenuation in the cavernous sinus. Subsequently, a CT angiogram of the head was obtained that showed opacification of the bilateral cavernous sinus and distention of the orbital veins (Figure 1A,B). Evaluation with diagnostic cerebral angiogram showed direct bilateral carotid-cavernous fistulas (CCFs) (Figures 2 and 3). She underwent staged embolization of the CCFs. Direct CCF is the abnormal connection between the internal carotid artery and the cavernous sinus.1, 2 CCFs can present with the classic triad of chemosis, pulsatile proptosis and ocular bruit.3, 4 Other symptoms include orbital pain, abducens nerve palsy, deterioration of visual acuity and ophthalmoplegia.4, 5 Causes of CCF formation are traumatic, iatrogenic, connective tissue disorders and spontaneous.2, 6-9 Traumatic CCFs formation occurs in up to 1.25% of patients with traumatic brain injury or facial and skull base fractures.10, 11 Bilateral CCF formation is rare after trauma and symptoms develop in a delayed fashion.10-15 Treatment options include conservative management, open surgery, endovascular procedures and radiosurgery.16-18

Direct to Angiosuite in Acute Stroke with Mobile Stroke Unit.

Early reperfusion has the best likelihood for a favorable outcome in acute ischemic stroke (AIS) with large vessel occlusion (LVO). Our experience with mobile stroke unit (MSU) for direct to angiosuite (DTAS) transfer in AIS patients with suspected LVO is presented. Retrospective review of prospectively collected data from November 2019 to August 2022, of patients evaluated and transferred by the University of Alberta Hospital MSU and moved to angiosuite for endovascular thrombectomy (EVT). A total of 41 cases were included. Nine were chosen for DTAS and 32 were shifted to angiosuite after stopping for computed tomography (CT) angiography of the head and neck (no-DTAS). Stroke severity measured by NIHSS (median with interquartile range (IQR)) was higher in patients of DTAS, 22 (14-24) vs 14.5 (5-25) in no-DTAS (p = 0.001). The non-contrast CT head in MSU showed hyperdense vessels in 8 (88.88%) DTAS vs 11 (34.35%) no-DTAS patients (p = 0.003). The EVT timelines (median with IQR, 90th percentile) including "door to artery puncture time" were 31 (23-50, 49.2) vs 79 (39-264, 112.8) minutes, and "door to recanalization time" was 69 (49-110, 93.2) vs 105.5 (52-178, 159.5) minutes in DTAS vs no-DTAS group, respectively. The workflow times were significantly shorter in the DTAS group (p < 0.001). Eight (88.88%) out of 9 DTAS patients had LVO and underwent thrombectomy. MSU for DTAS in patients with high NIHSS scores, cortical signs, and CT showing hyperdense vessel is an effective strategy to reduce the EVT workflow time.

Unified ICH quantification and prognosis prediction in NCCT images using a multi-task interpretable network.

With the recent development of deep learning, the regression, classification, and segmentation tasks of Computer-Aided Diagnosis (CAD) using Non-Contrast head Computed Tomography (NCCT) for spontaneous IntraCerebral Hematoma (ICH) have become popular in the field of emergency medicine. However, a few challenges such as time-consuming of ICH volume manual evaluation, excessive cost demanding patient-level predictions, and the requirement for high performance in both accuracy and interpretability remain. This paper proposes a multi-task framework consisting of upstream and downstream components to overcome these challenges. In the upstream, a weight-shared module is trained as a robust feature extractor that captures global features by performing multi-tasks (regression and classification). In the downstream, two heads are used for two different tasks (regression and classification). The final experimental results show that the multi-task framework has better performance than single-task framework. And it also reflects its good interpretability in the heatmap generated by Gradient-weighted Class Activation Mapping (Grad-CAM), which is a widely used model interpretation method, and will be presented in subsequent sections.

Pediatric low-dose head CT: Image quality improvement using iterative model reconstruction.

To evaluate the differences in pediatric non-contrast low-dose head computed tomography (CT) between filtered-back projection and iterative model reconstruction using objective and subjective image quality evaluation. A retrospective study evaluated children undergoing low-dose non-contrast head CT. All CT scans were reconstructed using both filtered-back projection and iterative model reconstruction. Objective image quality analysis was performed using contrast and signal-to-noise ratios for the supra- and infratentorial brain regions of identical regions of interest on the two reconstruction methods. Two experienced pediatric neuroradiologists evaluated subjective image quality, visibility of structures, and artifacts. We evaluated 233 low-dose brain CT scans of 148 pediatric patients. There was a ∼2-fold improvement in the contrast-to-noise ratio between gray and white matter in the infra- and supratentorial regions (p < 0.001) using iterative model reconstruction compared to filtered-back projection. The white and gray matter signal-to-noise ratio improved more than 2-fold using iterative model reconstruction (p < 0.001). Furthermore, radiologists graded anatomical details, gray-white matter differentiation, beam hardening artifacts, and image quality using iterative model reconstructions as superior to filtered-back projection reconstructions. Iterative model reconstructions had better contrast-to-noise and signal-to-noise ratios with fewer artifacts in pediatric CT brain scans using low-dose radiation protocols. This image quality improvement was demonstrated in the supra- and infratentorial regions. This method thus comprises an important tool for reducing children's exposure while maintaining diagnostic capability.

Prior knowledge-based precise diagnosis of blend sign from head computed tomography.

Automated diagnosis of intracranial hemorrhage on head computed tomography (CT) plays a decisive role in clinical management. This paper presents a prior knowledge-based precise diagnosis of blend sign network from head CT scans. We employ the object detection task as an auxiliary task in addition to the classification task, which could incorporate the hemorrhage location as prior knowledge into the detection framework. The auxiliary task could help the model pay more attention to the regions with hemorrhage, which is beneficial for distinguishing the blend sign. Furthermore, we propose a self-knowledge distillation strategy to deal with inaccuracy annotations. In the experiment, we retrospectively collected 1749 anonymous non-contrast head CT scans from the First Affiliated Hospital of China Medical University. The dataset contains three categories: no intracranial hemorrhage (non-ICH), normal intracranial hemorrhage (normal ICH), and blend sign. The experimental results demonstrate that our method performs better than other methods. Our method has the potential to assist less-experienced head CT interpreters, reduce radiologists' workload, and improve efficiency in natural clinical settings.