Keywords:
Liver, Artificial Intelligence, Oncology, CT, CT-Quantitative, Computer Applications-Detection, diagnosis, Computer Applications-General, Imaging sequences, Cancer, Metastases, Tissue characterisation
Authors:
M. Mottola1, A. Bevilacqua1, V. Vilgrain2; 1Bologna/IT, 2Clichy/FR
DOI:
10.26044/ecr2019/C-0522
Conclusion
This work employed 6 perfusion parameters and 4 statistical descriptors to compute 24 features,
to achieve 276 different couples analysed in order to select features permitting a linear separation into the bidimensional features space.
This feature selection analysis allowed selecting HPI as a promising feature to be applied in future steps of classification since its showed high performance in discriminating patients who will develop liver metastases,
with maximum specificity.
Results emphasized the potentiality of HPI in characterizing early vascular changes of livers,
which further developed metastases due to the reshaping of the blood supplying pathways.
As a general tendency,
the skewness of HPI for future metastatic patients is always associated to low values and centred around 0 a.u.
when coupled with BF,
BV,
MTT,
TTP,
thus suggesting that the separated patients hold normal distributions of HPI values.
Results strongly encourage the research on HPI-based biomarkers,
thus emphasizing the potential clinical role of CTp in treatments of tumours and hepatic diseases,
urging improving CTp standardization to speed up its introduction in the clinical routine.