A systematic review of the applications of Expert Systems (ES) and machine learning (ML) in clinical urology.

BACKGROUND: Testing a hypothesis for 'factors-outcome effect' is a common quest, but standard statistical regression analysis tools are rendered ineffective by data contaminated with too many noisy variables. Expert Systems (ES) can provide an alternative methodology in analysing data to identify variables with the highest correlation to the outcome. By applying their effective machine learning (ML) abilities, significant research time and costs can be saved. The study aims to systematically review the applications of ES in urological research and their methodological models for effective multi-variate analysis. Their domains, development and validity will be identified. METHODS: The PRISMA methodology was applied to formulate an effective method for data gathering and analysis. This study search included seven most relevant information sources: WEB OF SCIENCE, EMBASE, BIOSIS CITATION INDEX, SCOPUS, PUBMED, Google Scholar and MEDLINE. Eligible articles were included if they applied one of the known ML models for a clear urological research question involving multivariate analysis. Only articles with pertinent research methods in ES models were included. The analysed data included the system model, applications, input/output variables, target user, validation, and outcomes. Both ML models and the variable analysis were comparatively reported for each system. RESULTS: The search identified n = 1087 articles from all databases and n = 712 were eligible for examination against inclusion criteria. A total of 168 systems were finally included and systematically analysed demonstrating a recent increase in uptake of ES in academic urology in particular artificial neural networks with 31 systems. Most of the systems were applied in urological oncology (prostate cancer = 15, bladder cancer = 13) where diagnostic, prognostic and survival predictor markers were investigated. Due to the heterogeneity of models and their statistical tests, a meta-analysis was not feasible. CONCLUSION: ES utility offers an effective ML potential and their applications in research have demonstrated a valid model for multi-variate analysis. The complexity of their development can challenge their uptake in urological clinics whilst the limitation of the statistical tools in this domain has created a gap for further research studies. Integration of computer scientists in academic units has promoted the use of ES in clinical urological research.

Computed tomography-based psoas skeletal muscle area and radiodensity are poor sentinels for whole L3 skeletal muscle values.

BACKGROUND AND AIMS: Computed tomography (CT)-based measurement of skeletal muscle cross-sectional area (CSA) and Hounsfield unit (HU) radiodensity are used to assess the presence of sarcopenia and myosteatosis, respectively. The validated CT-based technique involves analysis of skeletal muscle at the third lumbar vertebral (L3) level. Recently there has been increasing interest in the use of psoas muscle alone as a sentinel. However, this technique has not been extensively investigated or compared with the previous validated standard approach. METHODS: Portovenous phase CT images at the L3 level were identified retrospectively from a single institution in 150 patients who had non-emergency scans and were analysed by a single assessor using SliceOmatic software v5.0 (TomoVision, Canada). Manual segmentation based upon validated HU thresholds for skeletal muscle density was performed for all skeletal muscle, as well as the individual muscle groups. The muscle CSA and mean radiodensity of each group were compared against the whole L3 slice values. RESULTS: When compared with whole L3 slice CSA, anterior abdominal wall CSA had the strongest correlation (r = 0.9315, p < 0.0001) followed by paravertebral (r = 0.8948, p < 0.0001), then psoas muscle (r = 0.7041, p < 0.0001). The mean ± SD density of the psoas muscle (42 ± 8.4 HU) was significantly higher than the whole slice radiodensity (32.3 ± 9.5 HU, p < 0.0001), with paravertebral radiodensity being a more accurate estimation (34.5 ± 10.8 HU). There was a significant difference in the prevalence of myosteatosis when the density measured from the psoas was compared with that of the whole L3 skeletal muscle (27.7% vs. 66.0%, p < 0.0001). CONCLUSION: Whole L3 slice CSA correlated positively with psoas muscle CSA but was subject to wide variability in results. Psoas muscle radiodensity was significantly greater than whole L3 slice density and resulted in underestimation of the prevalence of myosteatosis. Given the lack of equivalence from individual muscle groups, we recommend that further work be undertaken to investigate which muscle group, or indeed whether the gold standard of whole L3 skeletal muscle, provides the best correlation with clinical outcomes.

A comparison of two different software packages for analysis of body composition using computed tomography images.

OBJECTIVES: The analysis of body composition from computed tomography (CT) imaging has become widespread. However, the methodology used is far from established. Two main software packages are commonly used for body composition analysis, with results used interchangeably. However, the equivalence of these has not been well established. The aim of this study was to compare the results of body composition analysis performed using the two software packages to assess their equivalence. METHODS: Triphasic abdominal CT scans from 50 patients were analyzed for a range of body composition measures at the third lumbar vertebral level using OsiriX (v7.5.1, Pixmeo, Switzerland) and SliceOmatic (v5.0, TomoVision, Montreal, Canada) software packages. Measures analyzed were skeletal muscle index (SMI), fat mass (FM), fat-free mass (FFM), and mean skeletal muscle Hounsfield Units (SMHU). RESULTS: The overall mean SMI calculated using the two software packages was significantly different (SliceOmatic 51.33 versus OsiriX 53.77, P < 0.0001), and this difference remained significant for non-contrast and arterial scans. When FM and FFM were considered, again the results were significantly different (SliceOmatic 33.7 versus OsiriX 33.1 kg, P < 0.0001; SliceOmatic 52.1 versus OsiriX 54.2 kg, P < 0.0001, respectively), and this difference remained for all phases of CT. Finally, when analyzed, mean SMHU was also significantly different (SliceOmatic 32.7 versus OsiriX 33.1 HU, P = 0.046). CONCLUSIONS: All four body composition measures were statistically significantly different by the software package used for analysis; however, the clinical significance of these differences is doubtful. Nevertheless, the same software package should be used if serial measurements are being performed.

Symptomatic spinal epidural lipomatosis after combined hormonal and steroidal palliative therapy of prostate cancer.

INTRODUCTION: Spinal epidural lipomatosis (SLE) is an abnormal accumulation of unencapsulated fat. The association of chronic steroid therapy is described as well as obesity and Cushing's syndrome. SLE has a range of neurological presentations such as back pain, myelopathy, and cauda equina syndrome. Surgical decompression is known to improve symptomatic cases and prevent further deterioration. CASE PRESENTATION: An elderly male patient with background history of high-risk metastatic prostate cancer and long-term treatment with steroids and hormonal therapy presented with acute on chronic back pain and neurological symptoms. Spinal cord compression was excluded on the subsequent MRI. However, a spinal epidural lipomatosis causing significant central canal stenosis and compression of the cauda equine was revealed. DISCUSSION: With the increasing role of MRI in assessing back pain, asymptomatic spinal epidural lipomatosis has been frequently noticed. However, symptomatic presentation after long term combined treatment of steroids and hormonal therapy has not been reported in the literature.

Trans-arterial embolisation (TAE) in haemorrhagic pelvic injury: review of management and mid-term outcome of a major trauma centre.

BACKGROUND: Management of pelvic fracture associated haemorrhage is often complex with high morbidity and mortality rates. Different treatment options are used to control bleeding with an on-going discussion in the trauma community regarding the best management algorithm. MAIN BODY: Recent studies have shown trans-arterial embolisation (TAE) to be a safe and effective technique to control pelvic fracture associated haemorrhage. Computed tomography (CT) evidence of active bleeding, haemodynamic instability, and pelvic fracture patterns are amongst important indicators for TAE. CONCLUSION: Herein, we aim to provide a comprehensive literature review of the effectiveness of TAE in controlling haemorrhage secondary to pelvic fracture according to the indications, technique and embolic agents, and outcomes, whilst incorporating our Level 1 major trauma centre's (MTC) results between 2014-2017.

Body composition measurement using computed tomography: Does the phase of the scan matter?

OBJECTIVES: The aim of this study was to determine, from the methodologic standpoint, the effect of the presence or absence of intravenous contrast on body composition variables obtained by analysis of computed tomography (CT) images. METHODS: Triphasic abdominal (noncontrast, arterial phase, and portovenous phase contrast) CT scans from 111 patients were analyzed by two independent assessors at the third lumbar vertebral level using SliceOmatic software (version 5.0, TomoVision, Montreal, Canada). Variables included skeletal muscle index (SMI), fat and fat-free mass (FM and FFM, respectively), and mean skeletal muscle Hounsfield units (SMHU). RESULTS: Mean SMHU was lowest in the noncontrast phase (29.4, standard deviation [SD] 8.9 HU), followed by arterial (32.4, SD 9.3 HU) then portovenous phases (34.9, SD 9.4 HU). The mean skeletal muscle attenuation was significantly different depending on the phase of the scan in which the images were obtained. Calculated FM was significantly lower in both arterial (28.6, SD 8.8 kg, P < 0.0001) and portovenous phase scans (28.5, SD 8.9 kg, P < 0.0001) when compared with noncontrast (29.2, SD 8.9 kg). The mean FFM was not significantly different as measured on noncontrast, arterial, or portovenous phase CT scans (48, SD 11.2; 48.1, SD 9.8; and 48.6, SD 10.2 kg, respectively). No difference was seen in SMI. Interobserver reliability was high. CONCLUSIONS: The definition of myosteatosis should include a standardized phase of CT for analysis and this should be incorporated within its definition. However, as the magnitudes of the differences were relatively small, the effect of the phase of the scan on predicting outcome needs to be determined.

Incidental radiological diagnosis of asymptomatic Abernethy malformations-two case reports.

The diagnosis of the rare congenital extrahepatic portosystemic shunts is of clinical significance because of the risk of hepatic encephalopathy; liver dysfunction; and associated cardiac, gastrointestinal, vascular, skeletal and genitourinary anomalies. This article describes two varying cases showing the same type of the extrahepatic congenital shunts (Type II). Both the patients were clinically asymptomatic. The first patient initially presented with unprovoked deep venous thrombosis and a staging CT scan was performed to identify any potential underlying malignancy. The second was a polytrauma patient in whom a congenital extrahepatic portosystemic shunt was identified on the CT scan performed to investigate the trauma-related injuries. The first case underwent hepatological investigations, including a fibroscan to rule out liver fibrosis, and was diagnosed as having a Type II congenital malformation, while the second case is under observation post recovery from his traumatic injuries and will be subsequently referred to the hepatology team in the future. Although uncommon, extrahepatic portosystemic shunts can cause significant morbidity and mortality, and all new cases diagnosed radiologically should be further investigated by referring them to a hepatologist.

The impact of sarcopenia and myosteatosis on outcomes of unresectable pancreatic cancer or distal cholangiocarcinoma.

BACKGROUND & AIMS: Patients with pancreatic cancer have a poor prognosis, are often cachectic, and frequently demonstrate features of systemic inflammation, which may contribute to the phenomenon of myosteatosis. Analysis of body composition from CT scans has been used to study sarcopenia and its association with prognosis in a number of types of cancer, particular in combination with obesity. It has also been suggested that myosteatosis, defined as attenuated mean skeletal muscle Hounsfield units (HU), is associated with reduced survival in cancer. This study aimed to assess the association between body composition (sarcopenia and myosteatosis) and outcome in patients with unresectable pancreatic cancer. METHODS: All patients diagnosed with unresectable pancreatic cancer at Nottingham University Hospitals NHS Trust between 2006 and 2013 were considered for the study. A total of 228 patients were included retrospectively. Body composition was assessed using cross-sectional CT analysis to calculate a skeletal muscle index (SMI) for sarcopenia and use mean skeletal muscle HU for myosteatosis. RESULTS: The prevalence of sarcopenia in the whole patient group at baseline was 60.5% (138/228). Overall, patients who were sarcopenic had no significant difference in overall survival versus those who were not (p = 0.779). However, patients who were overweight/obese and sarcopenic had a significantly lower survival (p = 0.013). Of the 58 patients who were overweight or obese and sarcopenic, 32 were also myosteatotic. The prevalence of myosteatosis overall at baseline was 55.3% (126/228) and this was associated with significant reduction in overall survival (p = 0.049). Univariate Cox regression revealed myosteatosis but not sarcopenia to be predictive of reduced survival, however this relationship was lost on multivariate testing. Myosteatosis was associated with significantly greater levels of systemic inflammation (white cell count, neutrophil-lymphocyte ratio and C-reactive protein), anaemia and worsening of baseline blood urea. This relationship was not seen with sarcopenia. CONCLUSIONS: This is the largest study on the association between body composition and survival in patients with unresectable pancreatic cancer and has shown that although sarcopenia alone did not have a bearing on survival, the presence of myosteatosis was associated significantly with the presence of systemic inflammation and reduced survival.

Reliable EGFR mutation testing in ultrasound-guided supraclavicular lymph node fine-needle aspirates: a cohort study with diagnostic performance analysis.

INTRODUCTION: 15-30% of patients with lung cancer will have supraclavicular and cervical lymphadenopathy (SCLN). Ultrasound (US)-guided fine-needle aspiration (FNA) cytology is regarded as an effective diagnostic tool in small size lymph nodes (LNs) and impalpable positron emission tomography detected nodes. We evaluated our diagnostic service performance in relation to the adequacy of samples for epidermal growth factor receptor (EGFR) mutation. METHODS: Retrospective data analysis from electronic records, searching for all suspected lung cancer referrals that underwent US of the neck±FNA, over a continuous period of 4 years. RESULTS: Of 306 cases with suspected lung cancer referred to our department for US FNA of SCLN, 228 patients underwent the procedure. Of the remaining 78 patients, LNs were not detected in 52 cases and appeared benign in 26. Cytological diagnosis was established in 171 patients (75%) for treatment decisions without further investigations. The remaining 57 patients had further investigations; 45 reconfirmed the US-guided FNA diagnosis. The average LN size was 12.9 mm, and positive cytology was obtained in LNs ranging from 3 to 45 mm. Of 57 adenocarcinoma cases, 34 were tested for EGFR confirming 4 positive, 25 negative and 5 insufficient samples. No complications were recorded. CONCLUSIONS: US-guided FNA of SCLN remains an important diagnostic tool in lung cancer. Adequate tissue can be obtained for reliable diagnosis from LNs and for EGFR mutational analysis, without the need for more invasive and expensive investigations in more than 80% of cases.

Potential clinical implications of the "aortic collar" configuration of an anomalous left brachiocephalic vein.

We present a case report of an extremely rare congenital anomaly of the brachiocephalic vein in a previously healthy patient presenting with acute central pulmonary embolism. Clinical and imaging findings are discussed in the report along with several potential complications envisaged by the authors to inform clinicians about the importance of reporting this particular thoracic vascular variant.

Abdominal abscess due to retained gallstones 5 years after laparoscopic cholecystectomy.

A 76-year-old female patient with diabetes presented with pyrexia and a recurrent painful right sided loin swelling. One year previously she had undergone radiological drainage of a large right sided loin abscess. At index presentation she was investigated both radiologically and endoscopically and a source for the abscess was not found. On this presentation, a computed tomography scan confirmed a large retroperitoneal abscess pointing through the lateral abdominal wall musculature. Surgical drainage was undertaken whereby the abscess was drained and several large gallstones extruded through the incision. The patient subsequently recuperated and the wound has healed successfully by second intention. Five years previously the patient had undergone an "uncomplicated" laparoscopic cholecystectomy. This case highlights the catastrophic late effects of dropped gallstones during laparoscopic cholecystectomy.