The dissertation is devoted to the development of radiological differential diagnostic criteria for tuberculous (TS) and nonspecific spondylitis (NS), the creation of a prognostic diagnostic model for differential diagnosis based on the integrated use of radiological tools: radiography, digital tomosynthesis (DT), computed tomography (CT), magnetic resonance (MRI), and clinical data.
In the dissertation, data 140 adult patients who treated at the Ternopil Regional Tuberculosis Dispensary from 2004 to 2016, of which 95 patients had spondylitis (54 – TS, 41 – NS). A set of methods used in the work: laboratory, diagnostic imaging (radiography, DT, CT, MRI), histological and bacteriological studies, statistical and mathematical modeling.
The results of the study deepen the existing knowledge about radiological differential diagnosis of TS and NS. In the study, multifactorial analysis of signs of TS and NS was developed and applied of using complex diagnostic tool for the first time. The possibilities and diagnostic value of digital tomosynthesis was studied in the differential diagnosis of spondylitis. The effectiveness using a combination of MRI + CT and MRI + CT methods has been statistically proven. Indicators of the diagnostic value of the MRI method (sensitivity – 100%, specificity – 76.2%, accuracy – 93.6%) increased to 100%, 92.9% and 96.1% with the combined usage MRI and CT; and up to 100%, 83.3% and 93.8% – with the combined use MRI and DT.
In a comparative analysis of the results obtained, we identified important differences in the radiological semiotics of TS and NS, in particular, the bone-destructive process and paravertebral abscesses. Differential diagnostic criteria for TS (exudative-caseous and productive types) and NS, has been proposed. An original integrated prognostic model of spondylitis (Z-index) has been developed, which determines the probability of diagnosing TS. This model in the study demonstrated a high quality of the classifier (area under the ROC-curve 0.9263) and can be used for the differential diagnosis of TS and NS.
The combination of clinical and radiological features had the best prognostic value for differential diagnosis of TS and NS, which in our study increased the specificity and accuracy of diagnosis to 85% and 88% for CT (against 81.5% and 85.5%, respectively), 91.6 % and 94% for MRI (against 76.2% and 93.6%, respectively).
