We undertook a systematic review of the diagnostic accuracy of artificial intelligence-based software for identification of radiologic abnormalities ( computer-aided detection , or CAD) compatible with pulmonary tuberculosis on chest x-rays (CXRs). We searched four databases for articles published between January 2005-February 2019. We summarized data on CAD type, study design, and diagnostic accuracy. We assessed risk of bias with QUADAS-2. We included 53 of the 4712 articles reviewed: 40 focused on CAD design methods (“Development” studies) and 13 focused on evaluation of CAD (“Clinical” studies). Meta-analyses were not performed due to methodological differences. Development studies were more likely to use CXR databases with greater potential for bias as compared to Clinical studies. Areas under the receiver operating characteristic curve (median AUC [IQR]) were significantly higher: in Development studies AUC: 0.88 [0.82–0.90]) versus Clinical studies (0.75 [0.66–0.87]; p-value 0.004); and with deep-learning (0.91 [0.88–0.99]) versus machine-learning (0.82 [0.75–0.89]; p = 0.001). We conclude that CAD programs are promising, but the majority of work thus far has been on development rather than clinical evaluation. We provide concrete suggestions on what study design elements should be improved.

No abstract

Background. Quality gaps exist in the diagnostic evaluation of lung cancer patients. The initial CT chest guides the workup of patients with suspected lung cancer. We sought to determine how frequently CT reports provided guideline-concordant recommendations with regard to additional imaging studies and/or invasive diagnostic procedures. Methods. This was a retrospective study. The records of patients referred for investigation of suspected lung cancer between January 1, 2015, and June 30, 2016, were reviewed. Patients with confirmed lung cancer, for whom CT scan images and reports were available, are included. CT reports were reviewed, with attention to additional imaging studies and/or invasive diagnostic procedures suggested. These recommendations were examined against current guidelines for lung cancer diagnosis and staging, based on suspected disease stage. Results. One hundred forty-six patients are included in the analysis. Most patients were diagnosed with non-small-cell lung cancer (NSCLC), and 63% had advanced disease (stages III and IV). Only 12% of CT reports contained guideline-concordant recommendations for additional imaging studies, with PET scan suggested in only 6% of reports. Potential invasive diagnostic procedures were suggested in one fifth of CT reports, and only 58% of these recommendations were in keeping with current guidelines. In particular, transthoracic needle aspiration (TTNA) was suggested in 26% of patients despite advanced stage disease. Conclusion. Guideline-concordant recommendations for investigation of suspected lung cancer are rarely available on CT reports. This is true with respect to both imaging studies and invasive diagnostic procedures. Incorporation of more evidence-based suggestions may reduce quality gaps in lung cancer diagnosis and staging.

patients with non-surgical stage disease (n¼10); 2) imprinted cytological samples from positive mediastinoscopies during the intraoperative staging of patients with lung cancer (n¼11); 3) positive pleural fluid in patients with pulmonary nodule (n¼2). Then we performed FISH technique, evaluated the quality of the signal obtained, and compared the results with those obtained on paraffin sections. FISH technique on paraffin blocks was performed using 2XSSC/ proteinase K pretreatment as standardized by our lab. Cytology smears were destained and fixed in 10% methanol and incubated with FISH probe (ALK, ROS1 and MET). Result: All cytology cases had scorable signals and were easy to interpret. Also, as no pretreatment was required, assay time was shorter. Depending on cellularity, one same slide was useful for analysis of the three probes. When comparing with IHC and FISH studies, we obtained a 100% correlation with ALK (n¼23; positive¼2, negative¼21), ROS1 (n¼5, all negative) and MET (n¼5, all negative). Conclusion: This work allowed us to optimize the use of different cytology samples frequently available in advance stage NSCLC for FISH studies. The use of cytological material might improve turnaround time for results and can become a useful tool in pathology labs, in particular when paraffin included material is limited.