Evaluation of a neural network classifier for pancreatic masses based on CT findings

doi:10.1016/S0895-6111(97)00006-2

Computerized Medical Imaging and Graphics

Volume 21, Issue 3, May–June 1997, Pages 175-183

https://doi.org/10.1016/S0895-6111(97)00006-2 Get rights and content

Abstract

We have investigated a neural network classifier based on CT findings extracted by a radiologist for the differential diagnosis between the pancreatic ductal adenocarcinoma and mass-forming pancreatitis, and compared its classification performance with that of Bayesian analysis, Hayashi's quantification method II, and radiologists. The three computerized classification methods were designed to classify categorized CT findings extracted by a radiologist, and were trained and tested on 71 cases. There was comparable performance between the neural the network, the Bayesian analysis, Hayashi's quantification method II, and the radiologists, in classifying pancreatic carcinoma and inflammatory mass.

Reference (39)

RedelmeierD.A. et al.
Assessing predictive accuracy: how to compare Brier scores
J. Clin. Epidemiol.
(1991)
BergeronB.P. et al.
Data qualification: Logic analysis applied toward neural network training
Comput. Biol. Med.
(1994)
NeffC.C. et al.
Inflammatory pancreatic masses: problems in differentiating focal pancreatitis from carcinoma
Radiology
(1984)
FreenyP.C. et al.
Pancreatic ductal carcinoma: diagnosis and staging with dynamic CT
Radiology
(1988)
LuetmerP.H. et al.
Chronic pancreatitis: reassessment with current CT
Radiology
(1989)
DelMaschioA. et al.
Pancreatic cancer versus chronic pancreatitis: diagnosis with CA 19-9 assessment: US, CT, and CT-guided fine-needle biopsy
Radiology
(1991)
SchulteS.J. et al.
Root of the superior mesenteric artery in pancreatitis and pancreatic carcinoma: evaluation with CT
Radiology
(1991)
MaclinP.S. et al.
Using an artificial neural network to diagnose hepatic masses
J. Med. Syst.
(1992)
AsadaN. et al.
Potential usefulness of an artificial neural network for differential diagnosis of interstitial lung diseases: pilot study
Radiology
(1990)
BooneJ.M. et al.
Neural networks in radiologic diagnosis: I. Introduction and illustration
Invest. Radiol.
(1990)

GrossG.W. et al.

Neural networks in radiologic diagnosis: II. Interpretation of neonatal chest radiographs

Invest. Radiol.

(1990)

FujitaH. et al.

Application of artificial neural network to computer-aided diagnosis of coronary artery disease in myocardial SPECT bull's-eye images

J. Nucl. Med.

(1992)

MillerA.S. et al.

Review of neural network applications in medical imaging and signal processing

Med. Biol. Eng. Comput.

(1992)

KippenhanJ.S. et al.

Evaluation of a neural-network classifier for PET scans of normal and Alzheimer's disease subjects

J. Nucl. Med.

(1992)

ScottR.

Artificial intelligence: its use in medical diagnosis

J. Nucl. Med.

(1993)

TourassiG.D. et al.

Acute pulmonary embolism: artificial neural network approach for diagnosis

Radiology

(1993)

ScottJ.A. et al.

Neural network analysis of ventilation-perfusion lung scans

Radiology

(1993)

WuY. et al.

Artificial neural networks in mammography: application to decision making in the diagnosis of breast cancer

Radiology

(1993)

SieblerM. et al.

Real-time identification of cerebral microemboli with US feature detection by a neural network

Radiology

(1994)

Cited by (19)

The use of intelligent database systems in acute pancreatitis e A systematic review
2014, Pancreatology
Citation Excerpt :
Analysis of both amylase and lipase together did not significantly enhance the accuracy at 0.84 (95% CI: 0.79–0.89). Ikeda et al. used radiological findings extracted from a radiology image database to differentiate mass-forming pancreatitis from pancreatic ductal adenocarcinoma and found ANN to be comparable to experienced radiologists [15]. Severity prediction was better with ANNs than APACHE II [12,19,32], modified Glasgow score (GS) [12,19] and Ranson's criteria [19].
Acute pancreatitis (AP) is a complex disease with multiple aetiological factors, wide ranging severity, and multiple challenges to effective triage and management. Databases, data mining and machine learning algorithms (MLAs), including artificial neural networks (ANNs), may assist by storing and interpreting data from multiple sources, potentially improving clinical decision-making.
1) Identify database technologies used to store AP data, 2) collate and categorise variables stored in AP databases, 3) identify the MLA technologies, including ANNs, used to analyse AP data, and 4) identify clinical and non-clinical benefits and obstacles in establishing a national or international AP database.
Comprehensive systematic search of online reference databases. The predetermined inclusion criteria were all papers discussing 1) databases, 2) data mining or 3) MLAs, pertaining to AP, independently assessed by two reviewers with conflicts resolved by a third author.
Forty-three papers were included. Three data mining technologies and five ANN methodologies were reported in the literature. There were 187 collected variables identified. ANNs increase accuracy of severity prediction, one study showed ANNs had a sensitivity of 0.89 and specificity of 0.96 six hours after admission – compare APACHE II (cutoff score ≥8) with 0.80 and 0.85 respectively. Problems with databases were incomplete data, lack of clinical data, diagnostic reliability and missing clinical data.
This is the first systematic review examining the use of databases, MLAs and ANNs in the management of AP. The clinical benefits these technologies have over current systems and other advantages to adopting them are identified.
Diabetes mellitus in pancreatic cancer and the need for diagnosis of asymptomatic disease
2010, Pancreatology
Pancreatic cancer is strongly associated with the development of hyperglycemia, peripheral insulin resistance and diabetes mellitus, especially when presented as new-onset diabetes mellitus. Peripheral insulin resistance and hyperinsulinemia have been suggested to promote growth of pancreatic cancer cells, and therefore a relation between longstanding diabetes mellitus type 2 and pancreatic cancer has been implied. Epidemiological studies, though, give incongruent results to this problem. There are data supporting a tumor-derived influence on glucose metabolism, insulin secretion and eventually the development of diabetes mellitus in early stages of pancreatic cancer. The only possibility for curative intent in pancreatic cancer is to diagnose the disease before symptoms occur. Patients with newly diagnosed diabetes mellitus type 2 or hyperglycemia as a risk group have been recommended for primary screening for pancreatic cancer. To date, there is no specific biomarker to identify patients with an asymptomatic pancreatic cancer. The review discuss the relationship between pancreatic cancer and diabetes mellitus and the possibility of secondary screening of patients with newly diagnosed diabetes mellitus type 2 or hyperglycemia in an artificial neural network. PubMed was searched for articles using the Mesh term ‘pancreatic neoplasms’ combined with ‘insulin resistance’ and ‘glucose metabolism disorders’. Additional articles were retrieved through hyperlinks and by manually searching reference lists in original published articles. In total 36 articles were systematically reviewed.
Improving the classification of multiple disorders with problem decomposition
2006, Journal of Biomedical Informatics
Differential diagnosis of multiple disorders is a challenging problem in clinical medicine. According to the divide-and-conquer principle, this problem can be handled more effectively through decomposing it into a number of simpler sub-problems, each solved separately. We demonstrate the advantages of this approach using abductive network classifiers on the 6-class standard dermatology dataset. Three problem decomposition scenarios are investigated, including class decomposition and two hierarchical approaches based on clinical practice and class separability properties. Two-stage classification schemes based on hierarchical decomposition boost the classification accuracy from 91% for the single-classifier monolithic approach to 99%, matching the theoretical upper limit reported in the literature for the accuracy of classifying the dataset. Such models are also simpler, achieving up to 47% reduction in the number of input variables required, thus reducing the cost and improving the convenience of performing the medical diagnostic tests required. Automatic selection of only relevant inputs by the simpler abductive network models synthesized provides greater insight into the diagnosis problem and the diagnostic value of various disease markers. The problem decomposition approach helps plan more efficient diagnostic tests and provides improved support for the decision-making process. Findings are compared with established guidelines of clinical practice, results of data analysis, and outcomes of previous informatics-based studies on the dataset.
Artificial neural networks
2000, Surgery
Citation Excerpt :
They have demonstrated a superior diagnostic accuracy for the detection and classification of brain lesions49,50 and breast lesions51 on MRI. They also appear capable of evaluating pancreatic lesions on computed tomography52 and chronic liver disease on liver scintigrams.53 Similar to ANNs used for mammographic screening, computer-aided diagnosis of Papanicolaou smears54,55 and other pathological specimens—including breast fine-needle aspiration cytology specimens56,57—has obvious potential advantages.
Artificial intelligence in pancreatic cancer
2022, Theranostics
Implementation of the System of the Early Diagnostics of Pancreatic Cancer in Clinical Practice
2020, Proceedings - 2020 2nd International Conference on Control Systems, Mathematical Modeling, Automation and Energy Efficiency, SUMMA 2020

View all citing articles on Scopus

View full text

Evaluation of a neural network classifier for pancreatic masses based on CT findings

Abstract

J. Clin. Epidemiol.

Comput. Biol. Med.

Inflammatory pancreatic masses: problems in differentiating focal pancreatitis from carcinoma

Radiology

Pancreatic ductal carcinoma: diagnosis and staging with dynamic CT

Radiology

Chronic pancreatitis: reassessment with current CT

Radiology

Pancreatic cancer versus chronic pancreatitis: diagnosis with CA 19-9 assessment: US, CT, and CT-guided fine-needle biopsy

Radiology

Root of the superior mesenteric artery in pancreatitis and pancreatic carcinoma: evaluation with CT

Radiology

Using an artificial neural network to diagnose hepatic masses

J. Med. Syst.

Potential usefulness of an artificial neural network for differential diagnosis of interstitial lung diseases: pilot study

Radiology

Neural networks in radiologic diagnosis: I. Introduction and illustration

Invest. Radiol.

Neural networks in radiologic diagnosis: II. Interpretation of neonatal chest radiographs

Invest. Radiol.

Application of artificial neural network to computer-aided diagnosis of coronary artery disease in myocardial SPECT bull's-eye images

J. Nucl. Med.

Review of neural network applications in medical imaging and signal processing

Med. Biol. Eng. Comput.

Evaluation of a neural-network classifier for PET scans of normal and Alzheimer's disease subjects

J. Nucl. Med.

Artificial intelligence: its use in medical diagnosis

J. Nucl. Med.

Acute pulmonary embolism: artificial neural network approach for diagnosis

Radiology

Neural network analysis of ventilation-perfusion lung scans

Radiology

Artificial neural networks in mammography: application to decision making in the diagnosis of breast cancer

Radiology

Real-time identification of cerebral microemboli with US feature detection by a neural network

Radiology