Towards Robust Brain Midline Shift Detection: A YOLO-Based 3D Slicer Extension with a Novel Dataset.

Introduction Brain midline shift (MLS) is a life-threatening condition that requires urgent diagnosis and treatment. Aim of the Work The aim of this study was to assess brain midline shift using transcranial sonography in comparison to plain CT. Methods The study was carried out on 45 neurocritical patients of both sex admitted to the critical care department in Ain shams University Hospital. The ultrasound MLS was assessed through the temporal bone window by measuring the distance from skull to the third ventricle on both sides prior to plain CT. CT MLS was determined either by the distance from the external bone table to the center of the third ventricle bilaterally (method 1) or the distance between the ideal midline and the septum pellucidum (method 2). Results 60% of patients were males, the mean GCS on admission was 7.12±1.66.All the patients were mechanically ventilated. Measurement of MLS by transcranial sonography was possible in all 45 patients. MLS of 4.29 ± 2.17 mm. A MLS >6 mm was observed in 24% (11/45) of the patients. CT MLS was 5.18 ± 2.49 mm (using method 1) and 5.35 ± 2.64 mm (using method 2). A MLS >8 mm with CT was observed in 29% (13/45) of the patients. The sensitivity and the specificity of US to detect a significant MLS (that is, MLS >5 mm) was 100, 91.3% respectively when using CT method (2) to measure MLS. The sensitivity and the specificity of US to detect a significant MLS (that is, MLS >5 mm) was 95.2, 95.8% respectively when using CT method (1) to measure MLS. The correlation coefficient between (the difference between US MLS and CTMLS) and CTMLS was 0.619 (p < 0.001).The smaller the MLS the narrower the difference between the measurement of USMLS and CTMLS. The narrowest difference between US MLS and CT MLS 0.5 (0.5 -0.7) was at MLS <2 mm. At this reading was the most accurate point of comparison between US MLS and CTMLS. The relation between US MLS and GCS was statistically significant, the greater the US MLS the lower the GCS. The relation between US MLS and length of ICU stay and ventilation days was tested and it was statistically significant. The greater the US MLS the longer the length of ICU stay and ventilation days. There was significant relation between US MLS and mortality, 100 % of the cases survived with US MLS less than 4 mm. 100 % of the cases with US MLS greater than 6 mm died. The greater the US MLS the higher the mortality. Conclusion This study suggests that TCS can be used as an easy bedside tool to detect MLS in neurocritical patients.

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.

Midline shift (MLS), the amount of displacement of the brain's midline from its normal symmetric position due to illness or injury, is an important index for clinicians to assess the severity of traumatic brain injury (TBI). In this dissertation, an automated computer-aided midline shift estimation system is proposed. First, a CT slice selection algorithm (SSA) is designed to automatically select a subset of appropriate CT slices from a large number of raw images for MLS detection. Next, ideal midline detection is implemented based on skull bone anatomical features and global rotation assumptions. For the actual midline detection algorithm, a window selection algorithm (WSA) is applied first to confine the region of interest, then the variational level set method is used to segment the image and extract the ventricle contours. With a ventricle identification algorithm (VIA), the position of actual midline is detected based on the identified right and left lateral ventricle contours. Finally, the brain midline shift is calculated using the positions of detected ideal midline and actual midline. One of the important applications of midline shift in clinical medical decision making is to estimate the intracranial pressure (ICP). ICP monitoring is a standard procedure in the care of severe traumatic brain injury (TBI) patients. An automated ICP level prediction model based on machine learning method is proposed in this work. Multiple features, including midline shift, intracranial air cavities, ventricle size, texture patterns, and blood amount, are used in the ICP level prediction. Finally, the results are evaluated to assess the effectiveness of the proposed method in ICP level prediction.

The severity of traumatic brain injury (TBI) is known to be characterized by the shift of the middle line in brain as the ventricular system often changes in size and position, depending on the location of the original injury. In this thesis, the focus is given to processing of the CT (Computer Tomography) brain images to automatically calculate midline shift in pathological cases and use it to predict Intracranial Pressure (ICP). The midline shift measurement can be divided into three steps. First the ideal midline of the brain, i.e., the midline before injury, is found via a hierarchical search based on skull symmetry and tissue features. Second, the ventricular system is segmented from the brain CT slices. Third, the actual midline is estimated from the deformed ventricles by shape matching method. The horizontal shift in the ventricles is then calculated based on the ideal midline and the actual midline in TBI CT images. The proposed method presents accurate detection of the ideal midline using anatomical features in the skull, accurate segmentation of ventricles for actual midline estimation using the information of anatomical features with a spatial template derived from a magnetic resonance imaging (MRI) scan, and an accurate estimation of the actual midline based on the robust proposed multiple regions shape matching algorithm. After the midline shift is successively measured, features including midline shift, texture information of CT images, as well as other demographic information are used to predict ICP. Machine learning algorithms are used to model the relation between the ICP and the extracted features. By using systematic feature selection and parameter selection of the learning model, promising results on ICP prediction are achieved. The prediction results also indicate the reliability of the proposed midline shift estimation.

Abstracts from The 39^th Annual Symposium of the National Neurotrauma Society, including the AANS/CNS Joint Section on Neurotrauma and Critical Care

Deep learning-based prediction of mortality using brain midline shift and clinical information.

Minimally invasive surgery (MIS) for large supratentorial intracerebral hemorrhage (ICH) addresses primary brain injury by reducing mass effect, causing midline shift (MLS). We investigated the relationship between MIS, MLS, and functional outcomes in a large clinical trial. We conducted a post hoc analysis of patients with qualified computed tomography (CT) images from the phase 3 Minimally Invasive Surgery Plus Alteplase for Intracerebral Haemorrhage Evacuation trial. The exposure was MLS at the pineal gland (PG) and septum pellucidum (SP) on CT scans at diagnosis, end of treatment (EOT), and hospital discharge. The primary outcome was 30-day mortality. Secondary outcomes were mortality and modified Rankin scale score at day 365. We performed multivariable logistic regression and mediation analyses, adjusted for demographics, ICH characteristics, and treatment. Participants had a median age of 62years and a median ICH volume of 44mL, and 61% were male. Thirty-day mortality was 12%. MLS on EOT CT was significantly lower in surgical patients compared with medical patients. EOT MLS in day 30 surgical survivors was also significantly lower compared with that in medically treated survivors and nonsurvivors. The odds of 30-day mortality on adjusted analyses were significantly increased by 1mm in MLS at both the PG and SP (PG: odds ratio 1.22, 95% confidence interval 1.06-1.41; SP: odds ratio 1.22, 95% confidence interval 1.10-1.36). Thresholds of MLS change < 3mm (SP) and < 5mm (PG) were associated with mortality reduction. The association of MIS with 30-day mortality was mediated fully by change in either the SP or PG MLS from diagnostic to EOT CT. Change in MLS was significantly associated with one-year mortality and, for SP, with one-year good functional outcome (modified Rankin scale scores 0-3). MIS for ICH significantly reduces MLS. This reduction in MLS significantly mediates reduction in 30-day mortality with MIS and is associated with both short-term and long-term outcomes.

Introduction Severe traumatic brain injury (TBI), defined as head trauma associated with a Glasgow Coma Scale score of 3–8, is a major and challenging problem in critical care medicine. TBI has a broad spectrum of severity, pathology, physiology, and sequelae. Because of this, there are numerous ways in which to describe and classify TBI. Cerebral ischemia is considered the single most important secondary event affecting outcomes following severe TBI. Computed tomography (CT) is considered to be the gold standard to diagnose midline shift (MLS); serial CTs in neurosurgical ICU patients can be associated with significant morbidity and secondary brain injuries posttransfer related to their transport. Transcranial Doppler is a noninvasive method to measure cerebral blood flow velocity. It is a clinically useful tool in the diagnosis of complications that may occur in patients with TBI. Objective This work aimed to monitor the effect of hyperosmolar solutions on brain MLS by using transcranial ultrasonography (TCUS) in severe TBI. Patients and methods This study was an observational comparative prospective cohort study that was done in the Critical Care Department of Alexandria University Hospitals on 60 patients with severe TBI. The ultrasound MLS was measured through the temporal acoustic bone and was compared by two methods of assessment by CT. Results In the current study, there was a positive correlation between acute physiology and chronic health evaluation II and ICU stay and mortality. We also found that brain edema, MLS, and Glasgow Coma Score were improved after the use of hyperosmolar solution. The most important finding in this study was that TCUS could detect and monitor MLS with only a small difference in comparison to CT brain, so it provides a cheap, accurate noninvasive, and bedside tool for diagnosis and monitoring MLS. Conclusion This study suggests that TCUS is comparable to CT (the cornerstone tool in neuroimaging) in early diagnosis and follow-up of MLS in severe TBI after using hyperosmolar solutions.

Objective To discuss the risk factors for coagulopathy in acute isolated traumatic brain injury (TBI). Methods A retrospective study was performed on 191 patients with acute isolated TBI hospitalized from July 2012 to June 2015. There were 70 patients with coagulopathy (coagulopathy group) and 121 patients without coagulopathy (control group). Age, gender, injury type, midline shift on CT and injury severity (Glasgow Coma Scale, GCS) were analyzed to identify the independent risk factors for coagulopathy using the logistic regression analysis. Correlation between the independent risk factors and coagulation indices was analyzed. Results Injury severity, acute subdural hematoma, intraventricular bleeding and midline shift on CT were identified as the independent risk factors for coagulopathy(P 1 ). Furthermore, injury severity and acute subdural hematoma were respectively associated with abnormalities of international normalized ratio (INR) and fibrinogen (Fg) (P<0.05 or P<0.01), intraventricular bleeding with abnormalities of prothrombin time (PT) and platelet count (PC) (P<0.01), and midline shift on CT with abnormalities of Fg and PC (P<0.05). Conclusions Injury severity, acute subdural hematoma, intraventricular bleeding and midline shift on CT are independent risk factors for coagulopathy in patients with acute isolated TBI, and correlate with abnormalities of several coagulation indices. Changes in coagulation indices should be monitored accurately after TBI, and timely treatment of coagulopathy can improve the prognosis. Key words: Brain injuries; Blood coagulation disorders; Risk factors of coagulopathy

Deep Learning Model for the Differential Diagnosis of Nasal Polyps and Inverted Papilloma by CT Images: A Multicenter Study.

Objective: To evaluate the role of the degree of midline shift on CT scans in predicting clinical outcomes in traumatic brain injury (TBI).&#x0D; Materials and Methods: A prospective observational study was conducted at the Department of Neurosurgery of a tertiary care hospital. We included 148 patients. After fulfilling the inclusion criteria, the patient’s baseline data, including the patient's age, gender, and CT scan findings with the degree of midline shift, was noted. The patients were monitored for three months to evaluate the outcome. The collected data was analyzed using SPSS version 22.0.&#x0D; Results: Our study showed that 105 (70.9%) patients showed satisfactory outcomes while 43 (29.1%) showed unsatisfactory outcomes. Patients with no midline shift were 70, out of which 55 (78.6%) showed satisfactory outcomes. Similarly, patients with 1-2 mm midline shifts showed satisfactory outcomes in 39 (69.6%) while 3-5 mm midline shifts showed 11 (50%) satisfactory outcomes. In our study, the degree of brain midline shift on CT scan was a statistically significant outcome factor (p = 0.035).&#x0D; Conclusion: Patients with TBI who had an increasing degree of midline shift on brain CT scans had considerably worse clinical outcomes.&#x0D; Keywords: Midline shift, CT scan, Glasgow Coma Score, Head Injury.

Clinical characteristics and outcomes in patients with traumatic brain injury in China: a prospective, multicentre, longitudinal, observational study.

BackgroundThis study aims to determine the relationship between pupillary reactivity, midline shift and basal cistern effacement on brain computed tomography (CT) in moderate-to-severe traumatic brain injury (TBI). All are important diagnostic and prognostic measures, but their relationship is unclear.MethodsA total of 204 patients with moderate-to-severe TBI, documented pupillary reactivity, and archived neuroimaging were included. Extent of midline shift and basal cistern effacement were extracted from admission brain CT. Mean midline shift was calculated for each ordinal category of pupillary reactivity and basal cistern effacement. Sequential Chi-square analysis was used to calculate a threshold midline shift for pupillary abnormalities and basal cistern effacement. Univariable and multiple logistic regression analyses were performed.ResultsPupils were bilaterally reactive in 163 patients, unilaterally reactive in 24, and bilaterally unreactive in 17, with mean midline shift (mm) of 1.96, 3.75, and 2.56, respectively (p = 0.14). Basal cisterns were normal in 118 patients, compressed in 45, and absent in 41, with mean midline shift (mm) of 0.64, 2.97, and 5.93, respectively (p < 0.001). Sequential Chi-square analysis identified a threshold for abnormal pupils at a midline shift of 7–7.25 mm (p = 0.032), compressed basal cisterns at 2 mm (p < 0.001), and completely effaced basal cisterns at 7.5 mm (p < 0.001). Logistic regression revealed no association between midline shift and pupillary reactivity. With effaced basal cisterns, the odds ratio for normal pupils was 0.22 (95% CI 0.08–0.56; p = 0.0016) and for at least one unreactive pupil was 0.061 (95% CI 0.012–0.24; p < 0.001). Basal cistern effacement strongly predicted midline shift (OR 1.27; 95% CI 1.17–1.40; p < 0.001).ConclusionsBasal cistern effacement alone is associated with pupillary reactivity and is closely associated with midline shift. It may represent a uniquely useful neuroimaging marker to guide intervention in traumatic brain injury.

The Brain Trauma Foundation. The American Association of Neurological Surgeons. The Joint Section on Neurotrauma and Critical Care. Computed tomography scan features.

Midline shift (MLS) has been associated with unfavorable outcome in patients with intracerebral hemorrhage (ICH). However, the optimal criteria to define the MLS measurements that indicate future outcome in ICH patients are absent, and the quantitative threshold of MLS that differentiates favorable and poor clinical outcome should be further explored. We enrolled patients with ICH who underwent admission computed tomography (CT) within 6h after onset of symptoms. We assessed MLS at several locations, including the pineal gland, septum pellucidum, and cerebral falx. MLS(max) was defined as the maximum midline shift among these locations. Functional outcomes were assessed with the Modified Rankin Scale (mRS) at 3months. We performed multivariate logistic regression analysis to investigate the MLS locations for predicting poor outcome. ROC curve analysis was used to establish whether MLS values were predictive of 90-day poor outcome. In 199 patients with ICH, 78 (39.2%) patients had poor functional outcome at 3-month follow-up. Pineal gland shift, septum pellucidum shift, cerebral falx shift, and MLS(max) all showed a significant difference between poor outcome and favorable outcome (p<0.001). After adjustment for age, baseline Glasgow Coma Scale score, ICH location, time to initial CT, baseline ICH volume, and intraventricular hemorrhage, the MLS(max) was independently associated with poor outcome (p=0.032). MLS(max)>4mm (our proposed optimal threshold) was more likely to have poorer outcomes than those without (p<0.001). MLS(max) can be a good independent predictor of clinical outcome, and MLS(max)>4mm is an optimal threshold associated with poor outcome in patients with ICH.