Fusion of Moment Invariant Method and Deep Learning Algorithm for COVID-19 Classification

The COVID-19 pandemic has resulted in a global health crisis. The rapid spread of the virus has led to the infection of a significant population and millions of deaths worldwide. Therefore, the world is in urgent need of a fast and accurate COVID-19 screening. Numerous researchers have performed exceptionally well to design pioneering deep learning (DL) models for the automatic screening of COVID-19 based on computerised tomography (CT) scans; however, there is still a concern regarding the performance stability affected by tiny perturbations and structural changes in CT images. This paper proposes a fusion of a moment invariant (MI) method and a DL algorithm for feature extraction to address the instabilities in the existing COVID-19 classification models. The proposed method incorporates the MI-based features into the DL models using the cascade fusion method. It was found that the fusion of MI features with DL features has the potential to improve the sensitivity and accuracy of the COVID-19 classification. Based on the evaluation using the SARS-CoV-2 dataset, the fusion of VGG16 and Hu moments shows the best result with 90% sensitivity and 93% accuracy.

Positive predictive value of CT coronary angiography vs. CT fractional flow reserve in a real-world population

Background Computerised tomography coronary angiography and fractional flow reserve (CTCA and CT-FFR) are non-invasive diagnostic tools for the detection of flow limiting coronary artery stenoses. Although their negative predictive values are well established, there is a concern that the high sensitivity of these tests may lead to overestimation of coronary artery disease (CAD) and unnecessary invasive coronary angiography (ICA). We compared the positive predictive value (PPV) of CT-FFR with computerised tomography coronary angiography (CTCA) against the gold standard of ICA in different real-world patient groups. Methods We conducted a retrospective study of 477 patients referred for CTCA or CT-FFR for investigation of possible coronary ischaemia. Patients were excluded if the image quality was poor or inconclusive. Patient-based PPV was calculated to detect or rule out significant CAD, defined as more than 70% stenosis on ICA. A sub-analysis of PPV by indication for scan was also performed. Patients that underwent invasive non-hyperaemic pressure wire measurements had their iFR or RFR compared with their CT-FFR values. Results In a patient-based analysis, the overall PPV was 59.3% for CTCA and 76.2% for CT-FFR. This increased to 81.0% and 86.7% respectively for patients with stable angina symptoms. In patients with atypical angina symptoms, CT-FFR considerably outperformed CTCA with a PPV of 61.3% vs. 37.5%. There was not a linear relationship between invasive pressure wire measurement and CT-FFR value (r=0.23, p=0.265). Conclusion The PPV of CTCA and CT-FFR is lower in the real-world than in previously published trials, partly due to the heterogeneity of indication for the scan. However, in patients with typical angina symptoms, both are reliable diagnostic tools to determine the presence of clinically significant coronary stenoses. CT-FFR significantly outperforms CTCA in patients with more atypical symptoms and the targeted use of CT-FFR in this group may help to avoid unnecessary invasive procedures. FUNDunding Acknowledgement Type of funding sources: None.

EP.TH.756Post-operative sepsis: Is CT reliable in diagnosing the cause of post-operative sepsis?

Aim To assess the reliability of computerised tomography in post-operative sepsis in patients who previously underwent laparotomy in the same hospital admission. Introduction Method This study was done on re-operations following an abdominal operation from April 2012 to April 2016 at a district general hospital. Data regarding demography, initial operation, post-operative problems, pre-operative diagnosis, and use of computed tomography, the findings on CT, and operative findings were collected. A comparison was made between the CT findings and findings on re-laparotomy. Also noted was the number of patients who did not have a re-laparotomy following CT. Result There were 87 patients, of whom, 10 had no accessible notes and were removed from the list. Of the remaining 77, 53(68.8%) had CT scans, 24 did not (31.2%). Of these that had CT, 29 (54.7%) had findings which matched with intra-operative findings on re-laparotomy, in 12 (22.6%) CT findings did not match 4(7.5%) had minor findings not needing surgery and 8(15.5%) had normal or insignificant findings. Conclusions We found in this study that, CT did have a significant contributory role in the arsenal of methods to aid source control in those patients that are septic following an abdominal operation, and this should be used as a pre-operative diagnostic modality.

EP.TH.155Assessment of the appropriate use of imaging and reporting in acute pancreatitis in a secondary care setting

Background Acute pancreatitis is the world's most common gastrointestinal disease requiring hospital admission. Our audit aim was to assess the timeframe within which Abdominal Ultrasound Scans (USS) and Computerised Tomography (CT) were performed, reported indications for CT and the prognostic factors noted in imaging reports. Methodology A retrospective search included admissions with acute pancreatitis between 01/09/19 to 30/11/19, collecting demographics, admission time & date, time and date of both radiological investigations and imaging reports. Patients under the age of 18 were excluded. Results This search identified 75 patients (M:F, 40:35) with a median age of 53 (18-95) years. USS were performed within 24 hours of admission in 40.0% (n = 30) of cases. Out of the patients (n = 44) who received a CT scan, 15.9% (n = 7) were scanned after more than 72 hours of onset of symptoms and 84.1% (n = 37) were scanned within less than 72 hours. Furthermore, 88.6%(n = 39) of CT request indications were in keeping with our standards. The average length of hospital stay was 6.1 days when scanned within 72 hours and 11.8 days when scanned after more than 72 hours. Conclusion Only 17.3% (n = 13) of the patients audited met all our standards and were managed according to the guidelines. The average length of hospital stay was half in those who had a CT scan after more than 72 hours of admission and this was also used to monitor disease progression/regression.

Diagnostic value of tonsillectomy in positron emission tomography and computerised tomography negative carcinoma of unknown primary

<p class="abstract">Carcinoma of unknown primary (CUP) represents a heterogeneous group of malignancy metastasis unique biology of which remains poorly understood. Even after a complete diagnostic workup including positron emission tomography and computerised tomography (PET-CT) scans the primary site of origin may remain unclear. This case series analysed the diagnostic value of bilateral palatine tonsillectomy in evaluating PET-CT negative head and neck squamous cell CUP. From retrospective analysis of cluster of 68 cases with metastatic cervical nodes with no obvious primary referred for ENT evaluation during a period of 20 months, we identified 5 cases where pan endoscopy and radiological evaluation including PET-CT were negative in detecting the primary. All 5 patients underwent bilateral palatine tonsillectomy along with biopsies from base of tongue and nasopharynx as a part of completion of diagnostic work-up. Tonsillar squamous cell carcinoma was revealed by subsequent histopathology examination in 4 out of 5 patients despite negative PET-CT evaluation. Blind biopsies from other sites like nasopharynx and base of tongue were unfruitful. Our experience strongly emphasises the fact that bilateral palatine tonsillectomy has a high yield in detecting primary even in PET negative CUP, though it needs a larger evidence base.</p>