Technical report: 3D printing of the brain for use as a visual-aid tool to communicate MR imaging features of hypoxic ischaemic injury at term with non-physicians.

3D printing has been used in several medical applications. There are no reports however of 3D printing of the brain in children for demonstrating pathology to non-medical professionals such as lawyers. We printed 3D models of the paediatric brain from volumetric MRI in cases of severe and moderate hypoxic ischaemic injury as well as a normal age matched control, as follows: MRI DICOM data was converted to NifTI (Neuroimaging Informatics Technology Initiative) format; segmentation of the brain into CSF, grey, and white matter was performed; the segmented data was converted to STL format and printed on a commercially available scanner. The characteristic volume loss and surface features of hypoxic ischaemic injury are visible in these models, which could be of value in the communication of the nature and severity of such an insult in a court setting as they can be handled and viewed from up close.

Massive Mesenteric Lymphadenopathy Causing Protein-losing Enteropathy in Gaucher Disease.

Protein-losing enteropathy due to massive mesenteric lymphadenopathy is a rare complication of Gaucher disease which is generally refractory to treatment with enzyme replacement and substrate reduction therapies. It is postulated that lymph nodes may act as a "sanctuary site" into which these treatments cannot penetrate. We present the case of a male child with Gaucher disease who developed massive mesenteric lymph nodes despite otherwise successful treatment with enzyme replacement therapy, and subsequently developed protein-losing enteropathy. The sonographic and magnetic resonance appearances of this complication are shown. Large volume lymphadenopathy inevitably provokes concern about the possibility of malignancy, but in a patient with Gaucher disease-particularly with significant ascites and clinical features of protein-losing enteropathy-this rare complication should be considered.

Curved reformat of the paediatric brain MRI into a 'flat-earth map' - standardised method for demonstrating cortical surface atrophy resulting from hypoxic-ischaemic encephalopathy.

Hypoxic-ischaemic encephalopathy is optimally imaged with brain MRI in the neonatal period. However neuroimaging is often also performed later in childhood (e.g., when parents seek compensation in cases of alleged birth asphyxia). We describe a standardised technique for creating two curved reconstructions of the cortical surface to show the characteristic surface changes of hypoxic-ischaemic encephalopathy in children imaged after the neonatal period. The technique was applied for 10 cases of hypoxic-ischaemic encephalopathy and also for age-matched healthy children to assess the visibility of characteristic features of hypoxic-ischaemic encephalopathy. In the abnormal brains, fissural or sulcal widening was seen in all cases and ulegyria was identifiable in 7/10. These images could be used as a visual aid for communicating MRI findings to clinicians and other interested parties.

Accuracy of radiologists, nonradiologists, and laypeople for identifying children with cerebral cortical atrophy from "Mercator map" curved reconstructions of MRIs of the brain.

BACKGROUND: Using text reports to communicate bilateral, symmetric, and zonal cortical brain atrophy in children with term hypoxic ischemic injury (HII) to parents and legal professionals contesting compensation rights can be difficult. Using standard cross-sectional images for explaining bilateral, regional brain imaging to laypeople is also challenging. A single flattened image of the brain surface, much like a map of the earth is derived from a globe, can be generated from curved reconstruction of magnetic resonance imaging (MRI) scans, i.e., a Mercator map. Laypeople's ability to identify abnormal "Mercator brain maps," without prior training, requires evaluation before use in nonmedical settings. AIM: To determine the sensitivity and specificity of laypeople in detecting abnormal pediatric Mercator flat-earth maps of the brain, without prior training. METHODS AND MATERIALS: 10 Mercator brain maps were provided to 111 participants individually. The maps comprised 5 HII, 1 cortical dysplasia, and 4 normal cases. Participants were required to identify the abnormal scans. Sensitivity and specificity overall and for participants' subgroups were calculated. RESULTS: Overall sensitivity and specificity were 67% and 80%, respectively. General radiologists (n = 12) had sensitivity and specificity of 91.2% and 94.6%, respectively. Laypeople (n = 54) had a sensitivity of 67% and specificity of 80%. CONCLUSION: The high specificity and sensitivity of radiologists validated the technique for distinguishing abnormal scans, regarding cortical pathology. High specificity of laypeople for identifying abnormal brains using Mercator maps indicates that this is a viable communication tool for demonstrating cortical MRI abnormalities of HII in children to laypersons.

Artificial intelligence for interpretation of segments of whole body MRI in CNO: pilot study comparing radiologists versus machine learning algorithm.

BACKGROUND: To initiate the development of a machine learning algorithm capable of comparing segments of pre and post pamidronate whole body MRI scans to assess treatment response and to compare the results of this algorithm with the analysis of a panel of paediatric radiologists. METHODS: Whole body MRI of patients under the age of 16 diagnosed with CNO and treated with pamidronate at a tertiary referral paediatric hospital in United Kingdom between 2005 and 2017 were reviewed. Pre and post pamidronate images of the commonest sites of involvement (distal femur and proximal tibia) were manually selected (n = 45). A machine learning algorithm was developed and tested to assess treatment effectiveness by comparing pre and post pamidronate scans. The results of this algorithm were compared with the results of a panel of radiologists (ground truth). RESULTS: When tested initially the machine algorithm predicted 4/7 (57.1%) examples correctly in the multi class model, and 5/7 (71.4%) correctly in the binary group. However when compared to the ground truth, the machine model was able to classify only 33.3% of the samples correctly but had a sensitivity of 100% in detecting improvement or worsening of disease. CONCLUSION: The machine learning could detect new lesions or resolution of a lesion with good sensitivity but failed to classify stable disease accurately. However, further validation on larger datasets are required to improve the specificity and accuracy of the machine model.