Assessment of methods for volumetric quantification of intracerebral hematomas in computerized tomography images

Miguel Vera, Yoleidy Huérfano, Doris Barrera, Oscar Valbuena, Williams Salazar, María Isabel Vera, Carlos Hernández, Marisela Vivas, Maryury Borrero, Ángel Valentín Molina, Luis Javier Martínez, Juan Salazar, Elkin Gelvez, Yudith Contreras, Frank Sáenz


This work evaluates the performance of computational methodsaimed at volume generation of five intracerebral hematomas(ICH), present in multi-layer computed tomographyimages, by means of three complementary steps. First. Aground truth volume or reference volume (RV) is considered.This RV is obtained, by a neurosurgeon, using the manualplanimetric method (MPM). In a second step, the volumetryof the 5 ICH’s is obtained considering both the original versionof the ABC/2 method and two of its variants, identified inthis paper as ABC/3 method and 2ABC/3 method. The ABCmethods allow for calculating hematoma volume under thegeometric assumption that the ICH has an ellipsoidal shape.In a third step, a smart automatic technique (SAT) is implementedto generate the three-dimensional segmentation ofeach ICH. In the context of the present work, the expressionSAT method is used to refer to the new methodology proposedto calculate the volume of the ICH. In order to evaluatethe performance of the SAT, the Dice coefficient (Dc) is usedto compare the dilated segmentations of the ICH with the ICHsegmentations generated, manually, by a neurosurgeon. Finally,the percentage relative error is calculated as a measureto evaluate the methodologies considered. The results showthat the SAT method exhibits the best performance, generatingan average percentage error of less than 3%.


Texto completo: PDF


  • No hay Refbacks actualmente.

world map hits counter
web counter
web counter visitors by country counter
flag counter