Search

The objective of this work is to automatically produce radiological gradings of spinal lumbar MRIs and also localize the predicted pathologies. We show that this can be achieved via a Convolutional Neural Network (CNN) framework that takes intervertebral disc volumes as inputs and is trained only on disc-specific class labels. Our contributions are: (i) a CNN architecture that predicts multiple gradings at once, and we propose variants of the architecture including using 3D convolutions; (ii) showing that this architecture can be trained using a multi-task loss function without requiring segmentation level annotation; and (iii) a localization method that clearly shows pathological regions in the disc volumes. We compare three visualization methods for the localization. The network is applied to a large corpus of MRI T2 sagittal spinal MRIs (using a standard clinical scan protocol) acquired from multiple machines, and is used to automatically compute disk and vertebra gradings for each MRI. These are: Pfirrmann grading, disc narrowing, upper/lower endplate defects, upper/lower marrow changes, spondylolisthesis, and central canal stenosis. We report near human performances across the eight gradings, and also visualize the evidence for these gradings localized on the original scans. [ABSTRACT FROM AUTHOR]

The introduction of magnetic resonance imaging in spondyloarthritis constitutes the main improvement in imaging over the past two decades. X-rays remains the cornerstone of diagnosis, although magnetic resonance imaging is more sensitive in early stages of the disease. The primary advantage of magnetic resonance imagingI is its ability to visualize soft tissue inflammation and inflammatory lesions within bone in three dimensions. Studies continue to appear describing its diagnostic utility, its sensitivity in early disease, its reliability, its predictive value for the detection of damage progression and its value in the assessment of certain complications. In this paper we review the characteristics of this imaging technique and the advantages and limitations of magnetic resonance imaging in spondyloarthritis. [ABSTRACT FROM AUTHOR]

Abstract: Objective: To look for cerebral white matter MRI changes in patients with long-standing and disabling MS. Methods: We analyzed retrospectively brain MRIs (performed 10 or more years after symptom onset) of patients with MS diagnosis and expanded disability status scale of 6 or more. The following parameters were analyzed: total number of brain T2 hyperintensities; number of brainstem, cerebellar, corpus callosum, basal ganglia, and juxtacortical T2 hyperintensities; diffuse leukoencephalopathy score; total number of T1 hypointensities. Results: Eighty-four patients were included. The mean time between symptom onset and MRI was 20.2 years. Eight percent had less than 9 cerebral T2 hyperintensities. Posterior fossa, juxtacortical, and corpus callosum T2 hyperintensities, and T1 hypointensities lacked in respectively 19%, 12%, 47%, and 8%. Overall, normal MRI was not seen, 6% had abnormal MRI but did not meet Barkhof''s criteria, and the remaining 94% had MRI abnormalities fulfilling Barkhof''s criteria. Moderate or severe diffuse leukoencephalopathy was seen in 69%. Extensive diffuse leukoencephalopathy predominant to nodular lesions was seen in 5%. Conclusions: Despite long-standing and disabling MS, typical MRI abnormalities lacked in a minority of patients, and 6% did not fulfil Barkhof''s criteria. The majority showed moderate or severe diffuse leukoencephalopathy. [Copyright &y& Elsevier]

Abstract: Anorectal malformations (ARMs) are associated with a variety of spinal dysraphisms, of which clinical impact is often underestimated. A 6-year-old girl, with a history of rectovaginal fistula, presented with gait disturbance, asthenia, and worsening of fecal incontinence. The spinal magnetic resonance imaging (MRI) at 12 months of age had revealed a tethered cord and a little intradural lipoma. Within the lipoma, a small cystic lesion, interpreted as a cerebrospinal fluid (CSF)-filled loculation, was also described. A consecutive MRI showed a marked increase in size of the CSF-like cyst that was clearly hyperintense on diffusion-weighted imaging (DWI) and presented reduced apparent diffusion coefficient values (855 ± 109 s/mm2), not compatible with CSF values. This lesion, interpreted as an (epi)dermoid cyst, was removed and histologically confirmed; the spinal cord was untethered. The child''s lower limb motor deficit resolved rapidly after surgery, and the fecal incontinence slowly returned to the previous bowel habit. There is a growing interest in recognizing and defining spinal dysraphism in ARM patients because some abnormalities may carry severe clinical consequences. For this purpose, a standardized MR protocol is required, in which DWI plays a pivotal role to disclose associated dysembriogenetic lesions, in particular when a CSF loculation is detected. [Copyright &y& Elsevier]

The authors aimed to evaluate the magnetic resonance (MR) imaging features characterizing spontaneous intracranial hypotension (SIH) in comparison with radionucleotide cisternography (RNC) findings in this condition. Five patients with postural headache and idiopathic low cerebrospinal fluid (CSF) pressure presented to our institution from October 2000 to July 2002. RNC demonstrated early filling of the bladder in five cases, and CSF leak was detected on the left side of the cervicothoracic junction in four cases. In two cases, intracranial MR imaging failed to demonstrate typical diffuse pachymeningeal enhancement; however, spinal MR imaging showed intense epidural or meningeal enhancement in those cases. Post-contrast spinal MR imaging was performed in four patients, revealing intense epidural or meningeal enhancement, mainly at the ventral cervical level and dorsal thoracic level of the spine. Among the four patients with obvious CSF leaks detected on the RNC, three patients demonstrated slight dilatation of the perineural root sleeve, without epidural fluid collection or spinal hygroma. In SIH, spinal MR imaging was very helpful in establishing the diagnosis when intracranial MR imaging failed to demonstrate typical findings. However, identification of the level of a CSF leak using MRI alone remains difficult, since the variety of imaging results is considerable and typical findings do not necessarily appear. [Copyright &y& Elsevier]

Objectives: To report the elevation of the 14-3-3 protein and the complete denervation of hand muscles in idiopathic acute transverse myelitis (IATM) of the cervical cord. Case description: In a 29-year-old woman with a 2-week history of neck pain and repeated attenuated flus, subacute quadriplegia, hypaesthesia of both arms, a T3 sensory level, and urinary dysfunction occurred. Based upon the clinical findings, the cervical MRIs, and an elevated 14-3-3 protein in the CSF, IATM C4–C7 was diagnosed. Ten, 17, 28 and 61 days after onset, nerve conduction studies revealed complete denervation of the right abductor pollicis brevis and abductor digiti minimi muscles but gradual improvement of the compound muscle action potential of the left abductor pollicis brevis muscle. F-waves of the right median nerve were absent. Tibial somatosensory evoked potentials showed a prolonged central conduction time. Transcranial magnetic stimulation evoked a response in the left but not the right abductor digiti minimi muscle. Conclusion: IATM may cause elevation of the 14-3-3 protein and loss of motor axons originating from affected anterior horn cells. [Copyright &y& Elsevier]

Purpose: To assess the effect of prescription parameters on the dose received by the spine during palliative radiotherapy.Methods and Materials: In a survey, members of the Canadian Association of Radiation Oncologists were asked to define their prescription parameters for vertebral metastases. The depth of the spinal canal and vertebral body at 8 spinal levels was measured in 20 magnetic resonance imaging studies (MRIs). Survey results were applied to the measurements to assess the dose received at depth. The depth of spinal structures assessed at simulation and by diagnostic imaging was compared.Results: Prescriptions were most commonly to Dmax 3 cm or 5 cm using 60Co-6MV photons delivering 8–30 Gy in 1–10 fractions. Mean depths from MRI were: posterior spinal canal, 5.5 cm; anterior spinal canal, 6.9 cm; and anterior vertebral body, 9.6 cm. Application of the prescription parameters from the survey to these measurements showed a wide range in the dose at depth with variation in technique. Depths measured at simulation correlated well with diagnostic imaging.Conclusion: The spinal canal and vertebral body lie >5 cm beneath the skin, and the dose received varies by up to 50% with changes in prescription depth. We suggest a suitable prescription point for vertebral metastases and a method for determining this at simulation. [Copyright &y& Elsevier]