Automated Detection of Hydrocephalus in Pediatric Head Computed Tomography Using VGG 16 CNN Deep Learning Architecture and Based Automated Segmentation Workflow for Ventricular Volume Estimation

Sekkat H., Khallouqi A., El rhazouani O., Halimi A., Madkouri Y.

Madkouri Y., Sekkat H., El Merabet Y., Talbi M., Aggour M., Bentayeb F., Khallouqi A.

Sekkat H., Abdellah K.A., Madkouri Y., EL ATIFI W., Rhouch I., El Rhazouani O., Tahiri Z., Talbi M., Halimi A.

Abstract Accurate dosimetry in computed tomography (CT) is essential for patient safety and effective radiation management. This study presents the development of an automated algorithm designed to enhance patient dosimetry by facilitating size-specific dose estimates (SSDE) and organ dose estimations. Utilizing a Python-based script, the proposed method integrates advanced image preprocessing, contour detection, and mathematical calculations to quantify key metrics from CT images. This automated approach addresses the limitations of manual measurement techniques. A retrospective analysis was conducted on CT axial images from examinations acquired with an 80-detector scanner. The algorithm processes DICOM images, converts pixel values to Hounsfield Units, applies Gaussian smoothing, windowing, and thresholding, followed by morphological operations to refine segmentation. It measures the water equivalent diameter (Dw) and estimates both region SSDE and organ doses, incorporating tissue attenuation. Validation was performed using an adult anthropomorphic ATOM phantom, with organ doses measured by optically stimulated luminescence dosimeters. The results demonstrated the algorithm's potential in estimating SSDE and organ doses. Validation of the automated method revealed strong correlations for Dw and SSDE between the proposed method and manual measurements of five expert reviewers ranging from 0.86 to 0.99 for determination coefficient. Comparative analysis of organ doses showed close agreement between results from experimental setup against the proposed algorithm. The automated algorithm estimated brain dose with a mean of 21.8 mGy, while measurements from the ATOM phantom and CT Expo indicated 19.74 mGy and 23.05 mGy, respectively. For lung doses, the automated algorithm estimated 12.5 mGy compared to 11.0 mGy from the ATOM phantom and 13.1 mGy from CT Expo. Liver doses were measured at 12.7 mGy by the automated method, versus 12.1 mGy from the ATOM phantom and 11.1 mGy from CT Expo. This study shows the potential of automated image analysis techniques in enhancing dosimetry accuracy in CT examinations.

Khallouqi A., Sekkat H., Halimi A., El rhazouani O.

Huang K.T., McNulty J., Hussein H., Klinger N., Chua M.M., Ng P.R., Chalif J., Mehta N.H., Arnaout O.

While ventricular shunts are the main treatment for adult hydrocephalus, shunt malfunction remains a common problem that can be challenging to diagnose. Computer vision-derived algorithms present a potential solution. We designed a feasibility study to see if such an algorithm could automatically predict ventriculomegaly indicative of shunt failure in a real-life adult hydrocephalus population. We retrospectively identified a consecutive series of adult shunted hydrocephalus patients over an eight-year period. Associated computed tomography scans were extracted and each scan was reviewed by two investigators. A machine learning algorithm was trained to identify the lateral and third ventricles, and then applied to test scans. Results were compared to human performance using Sørensen–Dice coefficients, calculated total ventricular volumes, and ventriculomegaly as documented in the electronic medical record. 5610 axial images from 191 patients were included for final analysis, with 52 segments (13.6% of total data) reserved for testing. Algorithmic performance on the test group averaged a Dice score of 0.809 ± 0.094. Calculated total ventricular volumes did not differ significantly between computer-derived volumes and volumes marked by either the first reviewer or second reviewer (p > 0.05). Algorithm detection of ventriculomegaly was correct in all test cases and this correlated with correct prediction of need for shunt revision in 92.3% of test cases. Though development challenges remain, it is feasible to create automated algorithms that detect ventriculomegaly in adult hydrocephalus shunt malfunction with high reliability and accuracy.

Gebest M., Weiss C., Cho C.G., Hausner L., Froelich L., Foerster A., Santhanam N., Fontana J., Groden C., Wenz H., Maros M.E.

AbstractAutomated tools have been proposed to quantify brain volume for suspected dementia diagnoses. However, their robustness in longitudinal, real-life cohorts remains unexplored. We investigated if expert visual assessment (EVA) of atrophy progression is reflected by automated volumetric analyses (AVA) on sequential MR-imaging. We analyzed a random subset of 20 patients with two consecutive 3D T1-weighted examinations (median follow-up 4.0 years, LQ-UQ: 2.1-5.2, range: 0.2-10). Thirteen (65%) with cognitive decline, the remaining with other neuropsychiatric diseases. EVA was performed by two blinded neuroradiologists using a 3 or 5-point Likert scale for atrophy progression (scores ±0-2: no, probable and certain progression or decrease, respectively) in dementia-relevant brain regions (frontal-, parietal-, temporal lobes, hippocampi, ventricles). Differences of AVA-volumes were normalized to baseline (delta). Inter-rater agreement of EVA scores was excellent (κ=0.92). AVA-delta and EVA showed significant global associations for the right hippocampus (p=0.035), left temporal lobe (p=0.0092), ventricle volume (p=0.0091) and a weak association for the parietal lobe (p=0.067).Post hoctesting revealed a significant link for the left hippocampus (p=0.039). In conclusion, the associations between volumetric deltas and EVA of atrophy progression were robust for certain brain regions. However, AVA-deltas showed unexpected variance, and therefore should be used with caution in individual cases, especially when acquisition protocols vary.

Pyrgelis E., Velonakis G., Papageorgiou S.G., Stefanis L., Kapaki E., Constantinides V.C.

Idiopathic bormal pressure hydrocephalus (iNPH) is a neurological syndrome that clinically presents with Hakim’s triad, namely cognitive impairment, gait disturbances, and urinary incontinence. The fact that iNPH is potentially reversible makes its accurate and early diagnosis of paramount importance. Its main imaging characteristic is the dilation of the brain’s ventricular system and the imaging parameters are also included in its diagnostic criteria along with clinical data. There is a variety of different modalities used and a great number of imaging markers that have been described while assessing iNPH patients. The present literature review attempts to describe the most important of these imaging markers and to shed some light on their role in diagnosis, differential diagnosis, and possibly prognosis of this potentially reversible neurological syndrome.

Nagata M., Ichikawa Y., Domae K., Yoshikawa K., Kanii Y., Yamazaki A., Nagasawa N., Ishida M., Sakuma H.

The purpose is to evaluate whether deep learning-based denoising (DLD) algorithm provides sufficient image quality for abdominal computed tomography (CT) with a 30% reduction in radiation dose, compared to standard-dose CT reconstructed with conventional hybrid iterative reconstruction (IR). The subjects consisted of 50 patients who underwent abdominal CT with standard dose and reconstructed with hybrid IR (ASiR-V50%) and another 50 patients who underwent abdominal CT with approximately 30% less dose and reconstructed with ASiR-V50% and DLD at low-, medium- and high-strength (DLD-L, DLD-M and DLD-H, respectively). The standard deviation of attenuation in liver parenchyma was measured as image noise. Contrast-to-noise ratio (CNR) for portal vein on portal venous phase was calculated. Lesion conspicuity in 23 abdominal solid mass on the reduced-dose CT was rated on a 5-point scale: 0 (best) to −4 (markedly inferior). Compared with hybrid IR of standard-dose CT, DLD-H of reduced-dose CT provided significantly lower image noise (portal phase: 9.0 (interquartile range, 8.7–9.4) HU vs 12.0 (11.4–12.7) HU, P < 0.0001) and significantly higher CNR (median, 5.8 (4.4–7.4) vs 4.3 (3.3–5.3), P = 0.0019). As for DLD-M of reduced-dose CT, no significant difference was found in image noise and CNR compared to hybrid IR of standard-dose CT (P > 0.99). Lesion conspicuity scores for DLD-H and DLD-M were significantly better than hybrid IR (P < 0.05). Dynamic contrast-enhanced abdominal CT acquired with approximately 30% lower radiation dose and generated with the DLD algorithm exhibit lower image noise and higher CNR compared to standard-dose CT with hybrid IR.

Rub S.A., Alaiad A., Hmeidi I., Quwaider M., Alzoubi O.

Recent technological advancements, like big data analytics, is driving the growing adoption of cyber-physical systems and digital twins in the area of healthcare. Congenital hydrocephalus is one important example of recent healthcare data analytics. Congenital hydrocephalus is a buildup of excess cerebrospinal fluid (CSF) in the brain at birth. Congenital hydrocephalus can be lethal without treatment and represents an urgent issue in present-day clinical practice. Congenital hydrocephalus has a significant effect on a human entire life since it causes damage to the brain. It is important to accurately diagnose hydrocephalus early, which will help in the early treatment of the infant by a surgical procedure called ventriculoperitoneal (VP) shunt which will reduce the damage caused by hydrocephalus on the brain. Deep Learning is an evolving technology that is currently actively researched in the field of radiology. Compared to the traditional hydrocephalus diagnosing techniques, automatic diagnosing algorithms in deep learning can save diagnosis time, improve diagnosing accuracy, reduce cost, and reduce the radiologist's workload. In this paper, we have used a novel dataset collected from king Hussein medical center hospital in Jordan that consists of CT scans for hydrocephalus and non-hydrocephalus infants, the dataset has gone through multiple stages in preprocessing which are; cropping and filtering, normalization, segmentation (three segmentation techniques have been applied), and augmentation. These data have been used to build deep learning and machine learning models that will help physicians in the early and accurate diagnosing of congenital hydrocephalus which will lead to a decrease in the death rate and brain damage. The results of our models were impressive with a 98.5% accuracy for congenital hydrocephalus classification in infants' brain CT images.

El Mansouri M., Choukri A., Semghouli S., Talbi M., Eddaoui K., Saga Z.

Size-specific dose estimates (SSDE) are the latest topic of interest in patient radiation–dose studies in computed tomography (CT). The aim of this study is to calculate and evaluate the doses (SSDE) by measuring the effective diameter (ED) of cross-sectional images collected during CT examinations of the chest and abdomen in Moroccan hospitals. Doses (SSDE) were calculated based on cross-sectional images by measuring the effective diameters of 75 patients in both examinations (45 for the thorax and 30 for the abdomen). Specific conversion factors for (ED) were used to convert the registered CTDIvol to SSDE, according to the instruction in the American Association of Physicists (AAPM) Report 204. In thoracic CT, the CTDIvol and SSDE values ranged from 5.8 to 10.7 mGy (mean: 8.08) and 9.55 to 15.37 mGy (mean: 12.13), respectively. For abdominal CT, CTDIvol and SSDE values ranged from 4.8 to 12.2 mGy (mean: 7.95) and 8.01 to 14.15 mGy (mean: 11.31), respectively. The results show that the SSDE is a useful tool and could potentially educate CT operators on its effective use as a way to optimize radiation dose instead of CTDIvol, in particular to establish diagnostic reference levels.

Jha T.R., Quigley M.F., Mozaffari K., Lathia O., Hofmann K., Myseros J.S., Oluigbo C., Keating R.F.

Shunt malfunction is a common complication and often presents with hydrocephalus. While the diagnosis is often supported by radiographic studies, subtle changes in CSF volume may not be detectable on routine evaluation. The purpose of this study was to develop a novel automated volumetric software for evaluation of shunt failure in pediatric patients, especially in patients who may not manifest a significant change in their ventricular size. A single-institution retrospective review of shunted patients was conducted. Ventricular volume measurements were performed using manual and automated methods by three independent analysts. Manual measurements were produced using OsiriX software, whereas automated measurements were produced using the proprietary software. A p value < 0.05 was considered statistically significant. Twenty-two patients met the inclusion criteria (13 males, 9 females). Mean age of the cohort was 4.9 years (range 0.1–18 years). Average measured CSF volume was similar between the manual and automated methods (169.8 mL vs 172.5 mL, p = 0.56). However, the average time to generate results was significantly shorter with the automated algorithm compared to the manual method (2244 s vs 38.3 s, p < 0.01). In 3/5 symptomatic patients whose neuroimaging was interpreted as stable, the novel algorithm detected the otherwise radiographically undetectable CSF volume changes. The automated software accurately measures the ventricular volumes in pediatric patients with hydrocephalus. The application of this technology is valuable in patients who present clinically without obvious radiographic changes. Future studies with larger cohorts are needed to validate our preliminary findings and further assess the utility of this technology.

Huang Y., Moreno R., Malani R., Meng A., Swinburne N., Holodny A.I., Choi Y., Rusinek H., Golomb J.B., George A., Parra L.C., Young R.J.

In large clinical centers a small subset of patients present with hydrocephalus that requires surgical treatment. We aimed to develop a screening tool to detect such cases from the head MRI with performance comparable to neuroradiologists. We leveraged 496 clinical MRI exams collected retrospectively at a single clinical site from patients referred for any reason. This diagnostic dataset was enriched to have 259 hydrocephalus cases. A 3D convolutional neural network was trained on 16 manually segmented exams (ten hydrocephalus) and subsequently used to automatically segment the remaining 480 exams and extract volumetric anatomical features. A linear classifier of these features was trained on 240 exams to detect cases of hydrocephalus that required treatment with surgical intervention. Performance was compared to four neuroradiologists on the remaining 240 exams. Performance was also evaluated on a separate screening dataset of 451 exams collected from a routine clinical population to predict the consensus reading from four neuroradiologists using images alone. The pipeline was also tested on an external dataset of 31 exams from a 2nd clinical site. The most discriminant features were the Magnetic Resonance Hydrocephalic Index (MRHI), ventricle volume, and the ratio between ventricle and brain volume. At matching sensitivity, the specificity of the machine and the neuroradiologists did not show significant differences for detection of hydrocephalus on either dataset (proportions test, p > 0.05). ROC performance compared favorably with the state-of-the-art (AUC 0.90–0.96), and replicated in the external validation. Hydrocephalus cases requiring treatment can be detected automatically from MRI in a heterogeneous patient population based on quantitative characterization of brain anatomy with performance comparable to that of neuroradiologists.

Quon J.L., Han M., Kim L.H., Koran M.E., Chen L.C., Lee E.H., Wright J., Ramaswamy V., Lober R.M., Taylor M.D., Grant G.A., Cheshier S.H., Kestle J.R., Edwards M.S., Yeom K.W.

OBJECTIVEImaging evaluation of the cerebral ventricles is important for clinical decision-making in pediatric hydrocephalus. Although quantitative measurements of ventricular size, over time, can facilitate objective comparison, automated tools for calculating ventricular volume are not structured for clinical use. The authors aimed to develop a fully automated deep learning (DL) model for pediatric cerebral ventricle segmentation and volume calculation for widespread clinical implementation across multiple hospitals.METHODSThe study cohort consisted of 200 children with obstructive hydrocephalus from four pediatric hospitals, along with 199 controls. Manual ventricle segmentation and volume calculation values served as “ground truth” data. An encoder-decoder convolutional neural network architecture, in which T2-weighted MR images were used as input, automatically delineated the ventricles and output volumetric measurements. On a held-out test set, segmentation accuracy was assessed using the Dice similarity coefficient (0 to 1) and volume calculation was assessed using linear regression. Model generalizability was evaluated on an external MRI data set from a fifth hospital. The DL model performance was compared against FreeSurfer research segmentation software.RESULTSModel segmentation performed with an overall Dice score of 0.901 (0.946 in hydrocephalus, 0.856 in controls). The model generalized to external MR images from a fifth pediatric hospital with a Dice score of 0.926. The model was more accurate than FreeSurfer, with faster operating times (1.48 seconds per scan).CONCLUSIONSThe authors present a DL model for automatic ventricle segmentation and volume calculation that is more accurate and rapid than currently available methods. With near-immediate volumetric output and reliable performance across institutional scanner types, this model can be adapted to the real-time clinical evaluation of hydrocephalus and improve clinician workflow.

Ono K., Iwamoto Y., Chen Y., Nonaka M.

Neikter J., Agerskov S., Hellström P., Tullberg M., Starck G., Ziegelitz D., Farahmand D.