Aims and objectives
Hepatocellular carcinoma (HCC) is one of the malignancies with increase incidence in the last years,
the management being dependent on the stage of the underlying condition [1,2].
An early diagnosis assures best curative chances and good survival rates in the general population.
Artificial neural networks (ANN) are modular computerized diagnostic systems,
currently being developed for helping clinicians towards better assessing diseases (Figure 1) [3,4,5].
Our aim was to assess the role of artificial neural networks (ANN) for diagnose HCC cases and better prognosticate their evolution,...
Methods and materials
One-hundred and twelve (112) consecutive patients with focal liver lesions signed informed consents for entering the study,
between January 2011 and February 2014.
All patients underwent both spiral computer tomography (CT) and contrast enhanced ultrasound (CEUS) at the Imaging and Gastroenterology Department,
University of Medicine and Pharmacy Craiova.
Final diagnosis was based on post-treatment evaluation,
follow up and pathology,
when available.
Spiral CT data was obtained through automated segmenting the tumor and a tumor-free parenchyma portion in distinctive images and obtaining Gray Level Co-occurrence Matrix...
Results
We included 41 cases of HCC,
32 liver metastasis hypovascular (Mh) and hypervascular (MH),
16 hemangiomas (H) and 23 focal steatosis (FS).
The ANN network was previously trained,
this version receiving only imaging data; it was able to correctly identify each tumor type,
with an overall sensitivity of 97.3%,
specificity of 97.4%.
The PPV (predictive positive value) and the NPV (negative predictive value) were 98.6% and 95.06%,
respectively (Figure 6).
Conclusion
In conclusion,
we could successfully diagnose the malignancy in these patients by using an ANN system.
Therefore,
we believe that such tools may become worthy aids to clinical management of patients with various types of digestive pathologies,
being superior when employed in medical classification problems and improving the diagnostic accuracy with established quantitative parameters.
Personal information
Acknowledgement:This study was financed by Partnership project VIP SYSTEM (ID:2011–3,2-0503) and the author I.A.
Gheoneawas financed by POSDRU grant No.
159/1.5/S/136893.
References
El-Serag HB,
Rudolph L.
Hepatocellular carcinoma: epidemiology and molecular carcinogenesis.
Gastroenterology.
2007;132: 2557-2576.
El-Serag HB.
Hepatocellular Carcinoma N Engl J Med,
2011; 365:1118-1127.
Grossi E,
Mancini A,
Buscema M.
International experience on the use of artificial neural networks in gastroenterology.
Dig Liver Dis,
2007; 39:278-285.
Saftoiu A,
Vilmann P,
Gorunescu F,
et all.
Neural network analysis of dynamic sequences of EUS elastography used for the differential diagnosis of chronic pancreatitis and pancreatic cancer.
Gastrointestinal Endoscopy 2008;68:1086-94.
Săftoiu A,
Vilmann P,
Gorunescu F,
et all; European...