Computer-aided Diagnosis of Acute Abdominal Pain

This paper reports a controlled prospective unselected real-time comparison of human and computer-aided diagnosis in a series of 304 patients suffering from abdominal pain of acute onset.The computing system''s overall diagnostic accuracy (91·8%) was significantly higher than that of the most senior member of the clinical team to see each case (79·6%). It is suggested as a result of these studies that the provision of such a system to aid the clinician is both feasible in a real-time clinical setting, and likely to be of practical value, albeit in a small percentage of cases.     

Human and Computer-aided Diagnosis of Abdominal Pain: Further Report with Emphasis on Performance of Clinicians

This paper reports a controlled trial of human and computer-aided diagnosis in a series of 552 patients with acute abdominal pain. The overall diagnostic accuracy of the computer-aided system was 91·5% and that of the senior clinician to see each case was 81·2%. However, the clinician''s diagnostic performance improved markedly during the period of the trial. The proportion of appendices which perforated before operation fell from 36% to 4% during the trial, and the negative laparotomy rate dropped sharply. After the trial closed in August 1972 these figures reverted towards their pretrial levels.It is suggested that while computer-aided diagnosis is a valuable direct adjunct to the clinician dealing with the “acute abdomen,” he may also benefit in the short-term from the constant feedback he receives and from the disciplines and constraints involved in communicating with the computer.     

Simplified computer-aided diagnosis of acute abdominal pain.

A simplified version of a system for computer-aided diagnosis of acute abdominal pain has been tested by "new" personnel unfamiliar with the previous system. After a two-month learning period the system proved more accurate in its diagnoses than the unaided clinician, and during the first five months of using the system the unaided clinicians'' accuracy rose from 73% to 84%. When computer "feedback" was withdrawn the clinicians'' diagnostic accuracy reverted towards the previous, "unaided" level. These findings further validate the concept of the computer as a potentially valuable diagnostic aid but indicate that a training period and computer feedback are important factors in its 4 use.     

Computer-aided Diagnosis: Description of an Adaptable System,and Operational Experience with 2,034 Cases

This paper describes a system of computer-aided diagnosis using an English Electric KDF9 computer linked to a terminal in a busy clinical department. Data from a series of patients were recorded, coded, and entered into the computer, which then performed a Bayesian analysis and displayed diagnostic probabilities in an adaptable format. Experience in this setting suggests that computer diagnosis may be a valuable aid to the clinician.     

Prostate cancer identification: quantitative analysis of T2-weighted MR images based on a back propagation artificial neural network model

Computer-aided diagnosis (CAD) systems have been proposed to assist radiologists in making diagnostic decisions by providing helpful information. As one of the most important sequences in prostate magnetic resonance imaging (MRI), image features from T2-weighted images (T2WI) were extracted and evaluated for the diagnostic performances by using CAD. We extracted 12 quantitative image features from prostate T2-weighted MR images. The importance of each feature in cancer identification was compared in the peripheral zone (PZ) and central gland (CG), respectively. The performance of the computer-aided diagnosis system supported by an artificial neural network was tested. With computer-aided analysis of T2-weighted images, many characteristic features with different diagnostic capabilities can be extracted. We discovered most of the features (10/12) had significant difference (P<0.01) between PCa and non-PCa in the PZ, while only five features (sum average, minimum value, standard deviation, 10th percentile, and entropy) had significant difference in CG. CAD prediction by features from T2w images can reach high accuracy and specificity while maintaining acceptable sensitivity. The outcome is convictive and helpful in medical diagnosis.     

NIKEI: A New Inexpensive and Non-Invasive Scoring System to Exclude Advanced Fibrosis in Patients with NAFLD

Aims

To develop, validate and compare a non-invasive fibrosis scoring system for non-alcoholic fatty liver disease (NAFLD) derived from routinely obtained clinical and biochemical parameters.

Methods

267 consecutive patients with biopsy proven fatty liver or non-alcoholic steatohepatitis were randomly assigned to the estimation (2/3) or validation (1/3) group to develop a model for the prediction of advanced fibrosis. Univariate statistics were performed to compare patients with and without advanced fibrosis, and following a multivariate logistic regression analysis a new scoring system was constructed. This non-invasive Koeln-Essen-index (NIKEI) was validated and compared to the FIB-4 index by calculating the area under the receiver operating characteristic curve (AUC). We evaluated a stepwise combination of both scoring systems for the precise prediction of advanced fibrosis. To set in contrast, we additionally tested the diagnostic accuracy of the AST/ALT ratio, BARD score and the NAFLD fibrosis score in our cohort.

Results

Age, AST, AST/ALT ratio, and total bilirubin were identified as significant predictors of advanced fibrosis and used to construct the NIKEI with an AUC of 0.968 [0.937; 0.998] compared to 0.929 [0.869; 0.989] for the FIB-4 index. The absence of advanced fibrosis could be confirmed with excellent accuracy (99–100%). The positive predictive value of the FIB-4 index was higher (100% vs. 60%), however, the false negative rate was also high (33%). With a stepwise combination of both indices 82%–84% of biopsies would have been avoidable without a single misclassification. The AUROC for AST/ALT ratio, the NAFLD fibrosis score, and the BARD score were 0.81 (95% CI, 0.72–0.90), 0.96 (95% CI 0.92–0.99), and 0.67 (95% CI 0.55–0.78), respectively.

Conclusion

The NIKEI can reliably exclude advanced fibrosis in subjects with NAFLD. In combination with the FIB-4 index misclassification with inadequate clinical management can be avoided while the need for liver biopsies can be reduced.     

The endoscopic diagnosis of gastroesophageal malignancy. A cytologic review

Cytologic reports were compared to final diagnoses for 1,157 gastroesophageal samples from an eight-year period in order to evaluate the diagnostic accuracy of endoscopic cytology and to determine the significance of a "suspicious" cytologic report. In the subgroup of patients with adenocarcinoma evaluated by paired endoscopic biopsy and cytology, the relative and combined sensitivities of the sampling methods were studied. Cytologic examination was reported as positive or suspicious in 85% of 229 cases of malignancy. There were three false-positive diagnoses of squamous-cell carcinoma of the esophagus, representing 0.3% of all submitted samples. Suspicious cytologic reports were issued in 5% of all cases. The majority (63%) of patients with a suspicious cytologic report had a final diagnosis of malignancy, with gastric adenocarcinoma present in almost half of the cases. Adenocarcinoma was diagnosed in 168 of the patients. Combined endoscopic biopsy and cytology was more sensitive (96%) than biopsy alone (90%) in making the initial diagnosis. Cytology may be of particular value in the diagnosis of gastroesophageal malignancy when the lesions are small and superficial or where stricture precludes adequate biopsy. Regardless of the biopsy findings, patients with "suspicious" cytologic reports require careful reevaluation since a high percentage of those cases in our series were subsequently verified as having malignancy.     

人心脏型脂肪酸结合蛋白检测试剂盒的临床评价

目的评价人心脏型脂肪酸结合蛋白(H-FABP)检测试剂盒(胶体金法)在急性心肌梗死(AMI)诊断中的价值。方法采用平行、盲法、对照的对比试验设计,比较其试验产品和对比产品对诊断AMI的敏感性、特异性、准确性。结果共测定240份临床血液标本。试验产品和对比产品的阳性符合率为100%,阴性符合率为96.15%,总符合率为97.92%。对比产品和试验产品结果不一致的5例标本以临床诊断结果为标准进行验证后,试验产品与临床诊断结果的阳性符合率为100%,总符合率为100%。采用Kappa检验考核两种产品测定结果的一致性,Kappa指数为0.958。经McNamara's test分析,两产品之间无差异,χ2=3.20,P>0.05。结论试验产品显示出较好的诊断价值,可以作为AMI早期诊断标志物,试验产品与对比产品等效。     

Risk Models and Scoring Systems for Predicting the Prognosis in Critically Ill Cirrhotic Patients with Acute Kidney Injury: A Prospective Validation Study

Background

Cirrhotic patients with acute kidney injury (AKI) admitted to intensive care units (ICUs) show extremely high mortality rates. We have proposed the MBRS scoring system, which can be used for assessing patients on the day of admission to the ICU; this new system involves determination of mean arterial pressure (MAP) and bilirubin level and assessment of respiratory failure and sepsis. We had used this scoring system to analyze the prognosis of ICU cirrhotic patients with AKI in 2008, and the current study was an external validation of this scoring system.

Methods

A total of 190 cirrhotic patients with AKI were admitted to the ICU between March 2008 and February 2011. We prospectively analyzed and recorded the data for 31 demographic parameters and some clinical characteristic variables on day 1 of admission to the ICU; these variables were considered as predictors of mortality.

Results

The overall in-hospital mortality rate was 73.2% (139/190), and the 6-month mortality rate was 83.2% (158/190). Hepatitis B viral infection (43%) was observed to be the cause of liver disease in most of the patients. Multiple logistic regression analysis indicated that the MBRS and Acute Physiology and Chronic Health Evaluation III (ACPACHE III) scores determined on the first day of admission to the ICU were independent predictors of in-hospital mortality in patients. In the analysis of the area under the receiver operating characteristic (AUROC) curves, the MBRS scores showed good discrimination (AUROC: 0.863±0.032, p<0.001) in predicting in-hospital mortality.

Conclusion

On the basis of the results of this external validation, we conclude that the MBRS scoring system is a reproducible, simple, easy-to-apply evaluation tool that can increase the prediction accuracy of short-term prognosis in critically ill cirrhotic patients with AKI.     

Validating a Scoring System for the Diagnosis of Smear-Negative Pulmonary Tuberculosis in HIV-Infected Adults

Background

The challenge of diagnosing smear-negative pulmonary TB (tuberculosis) in people living with HIV justifies the use of instruments other than the smear test for diagnosing the disease. Considering the clinical-radiological similarities of TB amongst HIV-infected adults and children, the proposal of this study was to assess the accuracy of a scoring system used to diagnose smear-negative pulmonary TB in children and adolescents, in HIV-infected adults suspected of having smear-negative pulmonary TB.

Methods

A Phase III validation study aiming to assess the diagnostic accuracy of a scoring system for diagnosing smear-negative pulmonary TB in HIV-infected adults. The study assessed sensitivity, specificity, positive and negative likelihood ratios, and positive and negative predictive values of the scoring system. Three versions of the scoring system were tested.

Results

From a cohort of 2,382 (HIV-infected adults), 1276 were investigated and 128 were diagnosed with pulmonary TB. Variables associated with the diagnosis of TB were: coughing, weight loss, fever, malnutrition, chest X-ray, and positive tuberculin test. The best diagnostic performance occurred with the scoring system with new scores, with sensitivity = 81.2% (95%-CI 74.5% –88%), specificity = 78% (75.6% –80.4%), PPV = 29.2% (24.5% –33.9%) and NPV = 97.4% (96.4% –98.4%), LR+ = 3.7 (3.4–4.0) and LR− = 0.24 (0.2–0.4).

Conclusion

The proposed scoring system (with new scores) presented a good capacity for discriminating patients who did not have pulmonary TB, in the studied population. Further studies are necessary in order to validate it, thus permitting the assessment of its use in diagnosing smear-negative pulmonary TB in HIV-infected adults.     

Computer aided diagnosis of acute abdominal pain: a multicentre study.

A multicentre study of computer aided diagnosis for patients with acute abdominal pain was performed in eight centres with over 250 participating doctors and 16,737 patients. Performance in diagnosis and decision making was compared over two periods: a test period (when a small computer system was provided to aid diagnosis) and a baseline period (before the system was installed). The two periods were well matched for type of case and rate of accrual. The system proved reliable and was used in 75.1% of possible cases. User reaction was broadly favourable. During the test period improvements were noted in diagnosis, decision making, and patient outcome. Initial diagnostic accuracy rose from 45.6% to 65.3%. The negative laparotomy rate fell by almost half, as did the perforation rate among patients with appendicitis (from 23.7% to 11.5%). The bad management error rate fell from 0.9% to 0.2%, and the observed mortality fell by 22.0%. The savings made were estimated as amounting to 278 laparotomies and 8,516 bed nights during the trial period--equivalent throughout the National Health Service to annual savings in resources worth over 20m pounds and direct cost savings of over 5m pounds. Computer aided diagnosis is a useful system for improving diagnosis and encouraging better clinical practice.     

Automatic diagnosis of imbalanced ophthalmic images using a cost-sensitive deep convolutional neural network

Background

Ocular images play an essential role in ophthalmological diagnoses. Having an imbalanced dataset is an inevitable issue in automated ocular diseases diagnosis; the scarcity of positive samples always tends to result in the misdiagnosis of severe patients during the classification task. Exploring an effective computer-aided diagnostic method to deal with imbalanced ophthalmological dataset is crucial.

Methods

In this paper, we develop an effective cost-sensitive deep residual convolutional neural network (CS-ResCNN) classifier to diagnose ophthalmic diseases using retro-illumination images. First, the regions of interest (crystalline lens) are automatically identified via twice-applied Canny detection and Hough transformation. Then, the localized zones are fed into the CS-ResCNN to extract high-level features for subsequent use in automatic diagnosis. Second, the impacts of cost factors on the CS-ResCNN are further analyzed using a grid-search procedure to verify that our proposed system is robust and efficient.

Results

Qualitative analyses and quantitative experimental results demonstrate that our proposed method outperforms other conventional approaches and offers exceptional mean accuracy (92.24%), specificity (93.19%), sensitivity (89.66%) and AUC (97.11%) results. Moreover, the sensitivity of the CS-ResCNN is enhanced by over 13.6% compared to the native CNN method.

Conclusion

Our study provides a practical strategy for addressing imbalanced ophthalmological datasets and has the potential to be applied to other medical images. The developed and deployed CS-ResCNN could serve as computer-aided diagnosis software for ophthalmologists in clinical application.

     

Diagnostic accuracy and role of immediate interpretation of fine needle aspiration biopsy specimens from various sites

A statistical analysis of the accuracy of the immediate interpretation of 425 fine needle aspiration (FNA) biopsies from various sites was performed. Preliminary interpretation of Diff-Quik-stained smears was rendered within a few minutes after performing the FNA biopsy, using diagnostic terminology similar to that of a surgical pathology report. The immediate assessment was documented in the formal cytology report and compared to the final diagnosis. For the entire series, the immediate interpretation had a sensitivity of 96%, a specificity of 97%, a positive predictive value (PV) of 98%, a negative PV of 95% and an efficiency of the test of 96%. There were 14 false-negative or falsely insufficient immediate interpretations and one false-positive immediate diagnosis. The diagnostic accuracy of the immediate interpretation of FNA biopsies from specific sites was also calculated; FNA biopsies of the pancreas were the least accurate procedure, having a sensitivity of 60% but a specificity of 100%. The role of the immediate interpretation of FNA biopsies is similar to the use of frozen sections in surgical pathology. An immediate assessment can (1) determine whether an adequate specimen is present, (2) render a specific preliminary diagnosis, (3) guide further clinical investigations or treatment, and (4) determine whether ancillary studies are needed to make a more accurate or specific diagnosis from the FNA specimen. Our results indicate that the immediate interpretation of FNA biopsies is an accurate procedure that should be routinely employed.     

Computer-aided prediction of gangrenous and perforating appendicitis.

The clinical details of 100 patients with proved acute appendicitis were compared with those of 100 patients with perforating or gangrenous appendicitis. Twenty features were found to be significantly different between the two groups. This information was incorporated into a computer data base and used in the differential diagnosis of abdominal pain. A program written to predict the probability that gangrene or perforation was present in patients with appendicitis gave a diagnostic accuracy over 91%. A clinical scoring index, which accurately predicted the state of the appendix in 88% of patients, was constructed from the significant differences between the two groups. When clinical scoring or computer analysis predicts a high probability of perforation or gangrene in patients with appendicitis, surgery should be performed without delay.     

Predicting hospital-acquired infections by scoring system with simple parameters

Background

Hospital-acquired infections (HAI) are associated with increased attributable morbidity, mortality, prolonged hospitalization, and economic costs. A simple, reliable prediction model for HAI has great clinical relevance. The objective of this study is to develop a scoring system to predict HAI that was derived from Logistic Regression (LR) and validated by Artificial Neural Networks (ANN) simultaneously.

Methodology/Principal Findings

A total of 476 patients from all the 806 HAI inpatients were included for the study between 2004 and 2005. A sample of 1,376 non-HAI inpatients was randomly drawn from all the admitted patients in the same period of time as the control group. External validation of 2,500 patients was abstracted from another academic teaching center. Sixteen variables were extracted from the Electronic Health Records (EHR) and fed into ANN and LR models. With stepwise selection, the following seven variables were identified by LR models as statistically significant: Foley catheterization, central venous catheterization, arterial line, nasogastric tube, hemodialysis, stress ulcer prophylaxes and systemic glucocorticosteroids. Both ANN and LR models displayed excellent discrimination (area under the receiver operating characteristic curve [AUC]: 0.964 versus 0.969, p = 0.507) to identify infection in internal validation. During external validation, high AUC was obtained from both models (AUC: 0.850 versus 0.870, p = 0.447). The scoring system also performed extremely well in the internal (AUC: 0.965) and external (AUC: 0.871) validations.

Conclusions

We developed a scoring system to predict HAI with simple parameters validated with ANN and LR models. Armed with this scoring system, infectious disease specialists can more efficiently identify patients at high risk for HAI during hospitalization. Further, using parameters either by observation of medical devices used or data obtained from EHR also provided good prediction outcome that can be utilized in different clinical settings.     

"GOLD or lower limit of normal definition? a comparison with expert-based diagnosis of chronic obstructive pulmonary disease in a prospective cohort-study"

Background

The Global initiative for chronic Obstructive Lung Disease (GOLD) defines COPD as a fixed post-bronchodilator ratio of forced expiratory volume in 1 second and forced vital capacity (FEV1/FVC) below 0.7. Age-dependent cut-off values below the lower fifth percentile (LLN) of this ratio derived from the general population have been proposed as an alternative. We wanted to assess the diagnostic accuracy and prognostic capability of the GOLD and LLN definition when compared to an expert-based diagnosis.

Methods

In a prospective cohort study, 405 patients aged ≥ 65 years with a general practitioner''s diagnosis of COPD were recruited and followed up for 4.5 (median; quartiles 3.9; 5.1) years. Prevalence rates of COPD according to GOLD and three LLN definitions and diagnostic performance measurements were calculated. The reference standard was the diagnosis of COPD of an expert panel that used all available diagnostic information, including spirometry and bodyplethysmography.

Results

Compared to the expert panel diagnosis, ''GOLD-COPD'' misclassified 69 (28%) patients, and the three LLNs misclassified 114 (46%), 96 (39%), and 98 (40%) patients, respectively. The GOLD classification led to more false positives, the LLNs to more false negative diagnoses. The main predictors beyond the FEV1/FVC ratio for an expert diagnosis of COPD were the FEV1 % predicted, and the residual volume/total lung capacity ratio (RV/TLC). Adding FEV1 and RV/TLC to GOLD or LLN improved the diagnostic accuracy, resulting in a significant reduction of up to 50% of the number of misdiagnoses. The expert diagnosis of COPD better predicts exacerbations, hospitalizations and mortality than GOLD or LLN.

Conclusions

GOLD criteria over-diagnose COPD, while LLN definitions under-diagnose COPD in elderly patients as compared to an expert panel diagnosis. Incorporating FEV1 and RV/TLC into the GOLD-COPD or LLN-based definition brings both definitions closer to expert panel diagnosis of COPD, and to daily clinical practice.     

Diagnostic performance of multi-organ ultrasound with pocket-sized device in the management of acute dyspnea

Background

The availability of ultra-miniaturized pocket ultrasound devices (PUD) adds diagnostic power to the clinical examination. Information on accuracy of ultrasound with handheld units in immediate differential diagnosis in emergency department (ED) is poor. The aim of this study is to test the usefulness and accuracy of lung ultrasound (LUS) alone or combined with ultrasound of the heart and inferior vena cava (IVC) using a PUD for the differential diagnosis of acute dyspnea (AD).

Methods

We included 68 patients presenting to the ED of “Maurizio Bufalini” Hospital in Cesena (Italy) for AD. All patients underwent integrated ultrasound examination (IUE) of lung-heart-IVC, using PUD. The series was divided into patients with dyspnea of cardiac or non-cardiac origin. We used 2 × 2 contingency tables to analyze sensitivity, specificity, positive predictive value and negative predictive value of the three ultrasonic methods and their various combinations for the diagnosis of cardiogenic dyspnea (CD), comparing with the final diagnosis made by an independent emergency physician.

Results

LUS alone exhibited a good sensitivity (92.6%) and specificity (80.5%). The highest accuracy (90%) for the diagnosis of CD was obtained with the combination of LUS and one of the other two methods (heart or IVC).

Conclusions

The IUE with PUD is a useful extension of the clinical examination, can be readily available at the bedside or in ambulance, requires few minutes and has a reliable diagnostic discriminant ability in the setting of AD.

     

Clinical versus laboratory detection of alcohol abuse: the alcohol clinical index.

To determine reliable indicators of alcohol abuse a comprehensive set of clinical and laboratory information was acquired from three groups of subjects with a wide range of drinking histories: 131 outpatients with alcohol problems, 131 social drinkers, and 52 patients from family practice. Findings from clinical examination provided greater diagnostic accuracy than laboratory tests for detecting alcohol abuse. Logistic regression analysis produced an overall accuracy of 85-91% for clinical signs, 84-88% for items from the medical history, and 71-83% for laboratory tests in differentiating the three groups. Further analyses showed 17 clinical signs and 13 medical history items that formed a highly diagnostic instrument (alcohol clinical index) that could be used in clinical practice. A probability of alcohol abuse exceeding 0.90 was found if four or more clinical signs or four or more medical history items from the index were present. Despite recent emphasis on the laboratory diagnosis of alcohol abuse simple clinical measures seem to provide better diagnostic accuracy.     

Intraoperative cytology (squash smear) in neurosurgical practice – pitfalls in diagnosis experience based on 3057 samples from a single institution

OBJECTIVES: The smear technique is challenging for a neuropathologist where rapid and accurate diagnosis is to be given on small biopsies. The present study, a large retrospective analysis of squash smears in neurosurgical practice, was conducted to assess the usefulness, accuracy and the diagnostic pitfalls of smear diagnosis. METHODS: The authors analysed 3057 central nervous system (CNS) lesions sent for intraoperative cytology (IC) during the years 1988-2005. The stain used was 1% alcoholic toluidine blue. The smear diagnosis was compared with the histological diagnosis to evaluate the diagnostic accuracy. RESULTS: Diagnostic accuracy irrespective of lesion and site ranged from 83.0% to 86.0% per year (mean=85%). The highest rate of correlation among common brain tumours was noted in schwannoma (96.6%) and pituitary adenoma (92.2%), followed by meningiomas (88.9%), astrocytomas (88.4%), chordomas (86.4%) and neurocytomas (86.9%). Infections as a whole contributed 380 cases. The most common infection was tuberculosis. CONCLUSION: This is the largest series reported from India to the best of our knowledge. Squash smear technique is a very reliable and rapid method of intraoperative diagnosis. Knowledge of clinical and neuroimaging details helps the experienced neuropathologist to improve the diagnostic accuracy.