[Differential diagnosis of erosive osteochondrosis and bacterial spondylitis: magnetic resonance tomography (MRT)]

Differentiating Magnetic Resonance Images of Pyogenic Spondylitis and Spinal Modic Change Using a Convolutional Neural Network

STUDY DESIGN

A retrospective analysis of magnetic resonance imaging (MRI).

OBJECTIVE

The study aimed to evaluate the performance of a convolutional neural network (CNN) to differentiate spinal pyogenic spondylitis from Modic change on MRI. We compared the performance of CNN to that of four clinicians.

SUMMARY OF BACKGROUND DATA

Discrimination between pyogenic spondylitis and spinal Modic change is crucial in clinical practice. CNN deep-learning approaches for medical imaging are being increasingly utilized.

METHODS

We retrospectively reviewed MRIs from pyogenic spondylitis and spinal Modic change patients. There were 50 patients per group. Sagittal T1-weighted (T1WI), sagittal T2-weighted (T2WI), and short TI inversion recovery (STIR) MRIs were used for CNN training and validation. The deep learning framework Tensorflow was used to construct the CNN architecture. To evaluate CNN performance, we plotted the receiver operating characteristic curve and calculated the area under the curve. We compared the accuracy, sensitivity, and specificity of CNN diagnosis to that of a radiologist, spine surgeon, and two orthopedic surgeons.

RESULTS

The CNN-based area under the curves of the receiver operating characteristic curve from the T1WI, T2WI, and STIR were 0.95, 0.94, and 0.95, respectively. The accuracy of the CNN was significantly greater than that of the four clinicians on T1WI and STIR (P<0.05), and better than a radiologist and one orthopedic surgeon on the T2WI (P<0.05). The sensitivity was significantly better than that of the four clincians on T1WI and STIR (P<0.05), and better than a radiologist and one orthopedic surgeon on the T2WI (P<0.05). The specificity was significantly better than one orthopedic surgeon on T1WI and T2WI (P<0.05) and better than both orthopedic surgeons on STIR (P<0.05).

CONCLUSION

We differentiated between Modic changes and pyogenic spondylitis using a CNN that interprets MRI. The performance of the CNN was comparable to, or better than, that of the four clinicians.

Spine and Sacroiliac Joints Lesions on Magnetic Resonance Imaging in Early Axial-Spondyloarthritis During 24-Months Follow-Up (Italian Arm of SPACE Study)

Objectives: Our study aimed to identify: (1) the prevalence of spine and pelvis magnetic resonance imaging (MRI-spine and MRI-SIJ) inflammatory and structural lesions in patients (pts) with a diagnosis of axial spondyloarthritis (axSpA); (2) the predictive factors for a severe disease pattern with a higher probability of radiographic progression. Materials and Methods: Seventy-five pts with low back pain (LBP) (≥3 months, ≤2 years, onset ≤45 years) underwent physical examination, questionnaires, laboratory tests, X-rays, MRI-spine, and MRI-SIJ at baseline (T0) and during a 24-months follow-up. Two expert rheumatologists made axSpA diagnosis and classification (according ASAS criteria). MRI-spine, MRI-SIJ and X-rays were scored independently by 2 readers following the SPARCC, mSASSS, and mNY-criteria. According to ASAS criteria, 21 pts fulfilled imaging arm only and 29 clinical arm with/without imaging arm; 25 pts did not fulfill ASAS criteria. Results: At T0 the mean ± SD LBP onset was 28.51 ± 8.05 years, 45.3% pts were male, 38.7% were HLA-B27+; 56% showed bone marrow oedema (BMO) at MRI-spine and 64% at MRI-SIJ. Signs of enthesitis were found in 58% pts in the thoracic spine. Eighteen (24%) pts presented BMO at MRI-spine with a negative MRI-SIJ. The prevalence of BMO lesions and the SPARCC SIJ and spine score decreased during the follow-up in the 2 cohorts meeting ASAS criteria. An early onset of LBP, a lower use of NSAIDs, a BASDAI>4 were identified as predictors of spine structural damage; the high SPARCC SIJ score appeared to be a predictor of SIJ structural damage. A higher mSASSS was predicted by a lower age of onset of LBP. Predictor of higher SPARCC spine was a higher NSAIDs and of higher SPARCC SIJ score the HLA-B27 positivity with increased inflammatory biomarkers. Conclusions: At T0 a significant prevalence of BMO lesions was observed both in SIJ and spine, with predominant involvement of thoracic district. Since positive MRI-spine images were observed in the absence of sacroiliitis, these findings seem to be relevant in the axSpA diagnosis. Early age of disease onset, long duration of LBP, increased inflammatory biomarkers, higher use of NSAIDs, male gender, HLA-B27 positivity, SPARCC SIJ score>2 appeared predictors of radiological damage and activity.

Vertebral Bone Marrow and Endplate Assessment on MR Imaging for the Differentiation of Modic Type 1 Endplate Changes and Infectious Spondylodiscitis

On magnetic resonance (MR) imaging, Modic type 1 (MT1) endplate changes and infectious spondylodiscitis share similar findings. Therefore, this study investigated vertebral bone marrow and endplate changes to enable their differentiation. The lumbar spine MR examinations of 91 adult patients were retrospectively included: 39 with MT1; 19 with early spondylodiscitis without abscess; and 33 with advanced spondylodiscitis with abscess. The assessment included percentage of bone marrow edema on sagittal short tau inversion recovery images, and the signal ratio of edema to unaffected bone and endplate contour (normal; irregular, yet intact; blurred; destructive) on sagittal unenhanced T1-weighted images. Differences were tested for statistical significance by Chi-square test and mixed model analysis of variance. The MR diagnostic accuracy in differentiating MT1 and spondylodiscitis was assessed by cross-tabulation and receiver-operating characteristic analysis. The endplate contours, edema extents, and T1-signal ratios of MT1 (extent, 31.96%; ratio, 0.83) were significantly different (p < 0.001) from early spondylodiscitis (56.42%; 0.60), and advanced spondylodiscitis (91.84%; 0.61). The highest diagnostic accuracy (sensitivity, 94.87%; specificity, 94.23%; accuracy, 94.51%) in identifying MT1 was provided by an irregular, yet intact endplate contour. This may be a useful MR feature for the differentiation between MT1 and spondylodiscitis, particularly in its early stage.

Correlation between vacuum phenomenon on CT and fluid on MRI in degenerative disks

OBJECTIVE

The purpose of this study is to correlate the presence of intradiscal vacuum phenomenon on CT to that of intradiscal fluid on MRI.

SUBJECTS AND METHODS

In a prospective study, 20 patients with lumbar vacuum phenomenon on CT underwent two MRI examinations. One was performed after mobilization, and the other was performed after 6 hours of bed rest. T2-weighted turbo-spin echo (TSE), STIR, T1-weighted TSE, and four consecutive T2-weighted TSE sequences were performed on a 1.5-T scanner. Ninety-five discal segments were assessed for the presence of intradiscal fluid or hyperintense signal on the T2-weighted MRI examinations and were correlated with the presence of vacuum phenomenon on CT, degenerative endplate abnormalities on CT, and edema on MRI.

RESULTS

Sixty-nine of 95 discal segments (72.6%) showed vacuum phenomenon on CT. Sixteen of those 69 discal segments (23.1%) showed intradiscal fluid (n = 12) or hyperintense signal (n = 4) on MRI examinations performed after mobilization. Forty-one of 69 discal segments (59.4%) with vacuum phenomenon showed intradiscal fluid (n = 29) or hyperintense signal (n = 12) on MRI examinations performed after bed rest. Seventeen segments showed only fluid after bed rest. Nine segments showed more fluid after bed rest than after mobilization. Three segments showed an unchanged amount of fluid. There was a significant correlation between the presence of intradiscal fluid and the amount of bone marrow edema on MRI and the presence of degenerative endplate abnormalities on CT, respectively.

CONCLUSION

The replacement of intradiscal vacuum phenomenon by intradiscal fluid is a time- and position-dependent dynamic process and is related to Modic type 1 degenerative disk disease and degenerative endplate changes.

Diffusion-weighted imaging (DWI) in musculoskeletal MRI: a critical review

Magnetic resonance imaging (MRI) is the mainstay of diagnosis, staging and follow-up of much musculoskeletal pathology. Diffusion-weighted magnetic resonance imaging (DWI) is a recent addition to the MR sequences conventionally employed. DWI provides qualitative and quantitative functional information concerning the microscopic movements of water at the cellular level. A number of musculoskeletal disorders have been evaluated by DWI, including vertebral fractures, bone marrow infection, bone marrow malignancy, primary bone and soft tissue tumours; post-treatment follow-up has also been assessed. Differentiation between benign and malignant vertebral fractures by DWI and monitoring of therapy response have shown excellent results. However, in other pathologies, such as primary soft tissue tumours, DWI data have been inconclusive in some cases, contributing little additional information beyond that gained from conventional MR sequences. The aim of this article is to critically review the current literature on the contribution of DWI to musculoskeletal MRI.

Diffusion-weighted imaging (DWI) in musculoskeletal MRI: a critical review

Magnetic resonance imaging (MRI) is the mainstay of diagnosis, staging and follow-up of much musculoskeletal pathology. Diffusion-weighted magnetic resonance imaging (DWI) is a recent addition to the MR sequences conventionally employed. DWI provides qualitative and quantitative functional information concerning the microscopic movements of water at the cellular level. A number of musculoskeletal disorders have been evaluated by DWI, including vertebral fractures, bone marrow infection, bone marrow malignancy, primary bone and soft tissue tumours; post-treatment follow-up has also been assessed. Differentiation between benign and malignant vertebral fractures by DWI and monitoring of therapy response have shown excellent results. However, in other pathologies, such as primary soft tissue tumours, DWI data have been inconclusive in some cases, contributing little additional information beyond that gained from conventional MR sequences. The aim of this article is to critically review the current literature on the contribution of DWI to musculoskeletal MRI.

Whole-body MR imaging, bone diffusion imaging: how and why?

Whole-body MRI (W-B MRI) and diffusion-weighted imaging (DWI) are two novel techniques that greatly facilitate the evaluation of many disorders of childhood. In the musculoskeletal system, these techniques primarily aid in the evaluation of the marrow, although there is increasing interest in the study of soft-tissue abnormalities with W-B MRI and of cartilage with DWI.The normal pattern of marrow transformation affects both modalities throughout childhood. Haematopoietic marrow has a much higher signal intensity than fatty marrow on W-B MRI short tau inversion recovery (STIR) images (Darge et al. Eur J Radiol 68:289-298, 2008). Diffusion is greater in haematopoietic marrow than in fatty marrow and decreases in the skeleton with age (Jaramillo et al. Pediatr Radiol 34:S48, 2004). It is important therefore to remember that the entire skeleton is haematopoietic at birth and that there is a process of marrow transformation to fatty marrow. Marrow conversion proceeds from the fingers to the shoulders and from the toes to the hips. Within each bone, fatty marrow transformation begins in the epiphyses, and within the shaft of the long bones fatty marrow transformation begins at the diaphysis and proceeds towards the metaphyses.

Vertebral endplate signal changes (Modic change): a systematic literature review of prevalence and association with non-specific low back pain

The prevalence of "vertebral endplate signal changes" (VESC) and its association with low back pain (LBP) varies greatly between studies. This wide range in reported prevalence rates and associations with LBP could be explained by differences in the definitions of VESC, LBP, or study sample. The objectives of this systematic critical review were to investigate the current literature in relation to the prevalence of VESC (including Modic changes) and the association with non-specific low back pain (LBP). The MEDLINE, EMBASE, and SveMED databases were searched for the period 1984 to November 2007. Included were the articles that reported the prevalence of VESC in non-LBP, general, working, and clinical populations. Included were also articles that investigated the association between VESC and LBP. Articles on specific LBP conditions were excluded. A checklist including items related to the research questions and overall quality of the articles was used for data collection and quality assessment. The reported prevalence rates were studied in relation to mean age, gender, study sample, year of publication, country of study, and quality score. To estimate the association between VESC and LBP, 2 x 2 tables were created to calculate the exact odds ratio (OR) with 95% confidence intervals. Eighty-two study samples from 77 original articles were identified and included in the analysis. The median of the reported prevalence rates for any type of VESC was 43% in patients with non-specific LBP and/or sciatica and 6% in non-clinical populations. The prevalence was positively associated with age and was negatively associated with the overall quality of the studies. A positive association between VESC and non-specific LBP was found in seven of ten studies from the general, working, and clinical populations with ORs from 2.0 to 19.9. This systematic review shows that VESC is a common MRI-finding in patients with non-specific LBP and is associated with pain. However, it should be noted that VESC may be present in individuals without LBP.

Diffusion-weighted and diffusion tensor imaging for pediatric musculoskeletal disorders

Diffusion-weighted imaging (DWI) is a powerful tool that has recently been applied to evaluate several pediatric musculoskeletal disorders. DWI probes abnormalities of tissue structure by detecting microscopic changes in water mobility that develop when disease alters the organization of normal tissue. DWI provides tissue characterization at a cellular level beyond what is available with other imaging techniques, and can sometimes identify pathology before gross anatomic alterations manifest. These features of early detection and tissue characterization make DWI particularly appealing for probing diseases that affect the musculoskeletal system. This article focuses on the current and future applications of DWI in the musculoskeletal system, with particular attention paid to pediatric disorders. Although most of the applications are experimental, we have emphasized the current state of knowledge and the main research questions that need to be investigated.

Magnetic resonance imaging of spinal infection

This article reviews the pathophysiology of spinal infection and its relevance for imaging. Magnetic resonance imaging (MRI) is the modality with by far the best sensitivity and specificity for spinal infection. The imaging appearances of spinal infection in MRI are outlined, and imaging techniques are discussed. The problems of clinical diagnosis are outlined. There is some emphasis on the MRI differentiation of pyogenic and nonpyogenic infection and on the differential diagnosis of spinal infection centered on the imaging presentation.

Degenerative Erkrankungen der Wirbelsäule

GOAL

The aim of this article is to describe rare and often unrecognized causes of spinal pain syndromes.

METHOD

Intervertebral disc degeneration frequently appears in early adulthood and can have a symptomatic or asymptomatic course. This article discusses incidence, pathophysiology, imaging, and pain symptomatology involved in the origin of back pain.

RESULTS

Anulus tears are often found in asymptomatic individuals but could be implicated in lumbar pain symptomatology in correlation with the provocative discography. Transient disorders can lead to pseudarthrosis of the iliac bone and to degeneration or to a reactive hypermobility with intervertebral disc degeneration in the level above. Modic type 1 erosive osteochondrosis is characterized by bone marrow edema near the hyaline cartilage end plate, which mostly elicits severe pain and results in serious limitations in everyday activities. The most important differential diagnosis is spondylodiscitis. Schmorl's nodes can exhibit considerable surrounding bone marrow edema that can be mistaken for metastases. A combination of MRI and CT should be employed for the diagnostic work-up of fatigue fracture of the interarticular portion, which is often overlooked due to its location. Synovial cysts of the facet joints can lead to radicular symptoms. Insufficiency fracture of the sacrum is frequently mistaken for metastasis due to intense scintigraphic enhancement and its signal behavior in MRI. CT provides instructive information.

CONCLUSION

Differential diagnosis should include less common causes such as anulus tears, transient disorders, activated Schmorl's nodes, synovial cysts of the facet joints, fatigue fractures of the interarticular portion of the spine and the sacrum and distinguish from metastases in particular.

[Feasibilities and bounds of diagnostic radiology in case of back pain]

Chronic monotone back pain is no pressing indication for radiographic procedures, but chronic progressive or symptomatic back pain should be investigated by radiographic means. Beneath conventional radiology and computed tomography (CT) magnetic resonance imaging (MRI) has become a more method of standard in these cases. The radiographic investigation of back pain is shown in cases of discal and vertebral degeneration and spondylitis. Typical signs and differential diagnosis are demonstrated. After demonstration of radiological means. After introduction and valuation of radiological means, as conventional radiography, CT, MRI, myelography and scintigraphy, it is entered into degenerative changes and degenerative diseases of vertebra endplates and vertebra bodies as a reason of pain. Reasons of spinal stenosis are discussed. In case of inflammatory changes, bacterial inflammation of vertebrae and intervertebral joints are represented. Changes of spondylodiscitis/spondylitis are opposed to inflammatory changes of Morbus Bechterew and Morbus Scheuermann.

Diffusion weighted imaging of bone marrow pathologies

Diffusion weighted imaging of non-CNS tissue has attracted much attention during the last years. Its capability of probing the microstructure of a biologic tissue at a sub-millimeter range is used to evaluate its diffusion capacity, which is tissue specific and can be used for tissue characterization. Processes involving bone marrow where the primary target for DWI during the last years. Most experience has been gained for differentiating benign from pathologic vertebral compression fractures, which can be reliably done when quantitative diffusion measurements are available. However, preliminary results exist indicating that this non-invasive technique may be a potential tool for therapy monitoring, which will revise the management of cancer patients. Moreover, this will be the first non-invasive and quantifiable tool for evaluating the effectiveness of modern tumor treatment. In this article, we will give an overview on the current status of DWI in the evaluation of bone marrow alterations; on currently available DWI techniques and a short out-look on future aspects of DWI in bone marrow pathologies.

Magnetic resonance imaging of the axial skeleton in rheumatoid disease

The axial skeleton is a target for both spondyloarthritis and rheumatoid arthritis. While conventional radiography allows the clear documentation of the late stages of inflammatory changes, magnetic resonance imaging (MRI) is sensitive enough to depict early inflammatory lesions. It is, therefore, of particular importance for radiologists and clinicians to know the MRI appearances of inflammatory changes of the axial skeleton in rheumatoid diseases. Typical lesions in ankylosing spondylitis and related conditions comprise spondylitis (Romanus lesion), spondylodiscitis (Andersson lesion), arthritis of the apophyseal joints, the costovertebral and costotransverse joints, and insufficiency fractures of the ankylosed vertebral spine (non-inflammatory type of Andersson lesion). Sacroiliitis is associated with chronic changes such as sclerosis, erosions, transarticular bone bridges, periarticular accumulation of fatty tissue and ankylosis. In addition, acute findings include capsulitis, juxta-articular osteitis and the enhancement of the joint space after contrast medium administration. Another important sign of spondyloarthritis is enthesitis, which affects the interspinal and supraspinal ligaments of the vertebral spine and the interosseous ligaments in the retroarticular space of the sacroiliac joints. The main site of manifestation of spinal involvement in rheumatoid arthritis is the cervical spine. Typical changes are the destruction of the atlantoaxial complex by pannus tissue with subsequent atlantoaxial subluxation, basilar impression and erosion of the dens axis. Changes in the lower segments of the cervical spine are destruction of the apophyseal joints resulting in the so-called stepladder phenomenon. Because of the uniform response of the discovertebral complex to different noxae, a number of different conditions must be distinguished on the basis of the patient's clinical findings and history in combination with their imaging appearance. These conditions comprise degenerative disc disease, septic spondylodiscitis, Scheuermann's disease, Paget's disease and diffuse idiopathic skeletal hyperostosis (DISH).

[Osteomyelitis of the spine]

Spinal infections are rare, occurring most often in elderly patients with urinary tract infections or diabetes. With the increasing number of patients with immune suppression, and also the increasing number of immigrants in the population, spinal infections are seen more frequently, especially in young adults. Typically spinal infections are monomicrobial, Staphylococcus aureus being the most common organism. Hematogenous spread of bacteria through the arterial paravertebral collateral vessels into the subchondral bone marrow of the vertebral bodies is the most common source of infection. Clinical presentation is often nonspecific. Important diagnostic measurements are laboratory studies, radiological evaluation including MR image scans, and CT-guided percutaneous biopsy of the lesion for microbiological studies. The management of spinal infections consists of antimicrobial therapy over 6-8 weeks. Surgical intervention is indicated in neurologically compromised patients for spinal instability and abscesses.

Diffusion-weighted imaging of bone marrow: current status

Diffusion-weighted imaging allows for measurement of tissue microstructure and reflects the random motion of water protons. It provides a new method to study bone marrow and bone marrow alterations on the basis of altered water-proton mobility in various diseases. Different diffusion-weighted methods have proved to be capable of differentiating between benign edema and tumorous involvement of bone marrow. It is especially useful for the distinction of acute benign osteoporotic and malignant vertebral compression fractures. Diagnosis is based on the contrast to normal bone marrow. Hypo- or isointensity reflects acute benign collapse, whereas hyperintensity is indicative of the tumorous nature of a fracture. Apparent diffusion coefficients (ADC) are significantly lower in metastatic disease than in bone marrow edema. Furthermore, bone marrow cellularity can be estimated by ADC measurements. Diffusion-weighted imaging might be helpful for monitoring response to therapy in metastatic disease.

FDG positron emission tomography for differentiation of degenerative and infectious endplate abnormalities in the lumbar spine detected on MR imaging

OBJECTIVE

The objective of our study was to evaluate the usefulness of FDG positron emission tomography (PET) for the differentiation of degenerative and infectious endplate abnormalities in the lumbar spine that were detected on MR imaging.

SUBJECTS AND METHODS

FDG PET was performed prospectively in 30 consecutive patients with substantial endplate abnormalities (craniocaudal diameter of bone marrow abnormalities, > or = 25% of vertebral height) found during MR imaging of the lumbar spine. Both the MR and PET images were evaluated by two experienced musculoskeletal radiologists and two experienced nuclear physicians. The diagnosis of either degeneration with different types of endplate abnormalities or disk-space infection was determined. Clinical follow-up and, in selected cases, bone biopsies with cultures were used as the standard of reference.

RESULTS

On the MR images, 25 of the 38 degenerated levels were classified as Modic type I, 13 levels as type II, and none as type III. Five disk-space infections were diagnosed in four patients. MR imaging findings were false-positive at one disk level with type I abnormalities and false-negative at two levels with infection. PET did not show FDG uptake in the intervertebral spaces of any patient with degenerative disease. FDG PET findings were true-positive in all five levels with disk-space infection. The sensitivity and specificity for MR imaging in detecting disk-space infection were 50% and 96%, and were 100% and 100% for FDG PET, respectively (not significant, McNemar test, p = 0.5).

CONCLUSION

Our findings suggest that FDG PET may prove useful for differentiation of degenerative and infectious endplate abnormalities detected on MR imaging. Even in active (Modic type I) degenerative endplate abnormalities in our series, PET did not show increased FDG uptake.

Imaging of spinal infection

MR imaging is the modality of choice for the detection, staging, and differential diagnosis of inflammatory disorders of the spine. Infectious spondylitis is characterized by the involvement of two adjacent vertebrae and the intervening disk with severe BME and early destruction of the end plates. The disk space is narrowed and typically exhibits water-equivalent signal intensity on T2-weighted or STIR images. Prevertebral and epidural extensions, abscess formation, enhancement of the BME, the disk space, and the surrounding granulation tissue are well demonstrated by gadolinium-enhanced images. Cervical spondylitis frequently involves more than one level. Bone marrow abnormalities may be subtle at this level and increased signal intensity of the disk space on T2-weighted or STIR images is an important finding. The risk for neurologic complications is increased. Granulomatous infections caused by tuberculosis, brucellosis, fungi, and parasites, including hydatid disease (Echinococcus), are frequently associated with imaging findings different from those seen with nonspecific bacterial infection. In patients with chronic infectious spondylitis, diffuse reactive bone marrow changes with decreased signal intensity on T1-weighted images, increased signal intensity on T2-weighted and STIR images, and increased uptake after gadolinium administration may occur. This phenomenon is probably caused by reactive bone marrow stimulation, simulating diffuse hematologic neoplastic disease. Erosive intervertebral osteochondrosis with bandlike disk gadolinium enhancement and BME, which is commonly associated with local pain, is the most important differential diagnosis of bacterial spondylitis.