Diffusion tensor imaging for intracerebral hemorrhage outcome prediction: comparison using data from the corona radiata/internal capsule and the cerebral peduncle

Tetsuo Koyama,Masao Tsuji,Hiroyuki Nishimura,Hiroji Miyake,Takehisa Ohmura,Kazuhisa Domen
DOI: https://doi.org/10.1016/j.jstrokecerebrovasdis.2011.06.014
Abstract:Background: Magnetic resonance-diffusion tensor imaging (DTI) was used to predict motor outcome for patients with intracerebral hemorrhage. We compared the predictive accuracy of data sampled from the cerebral peduncle with data from the corona radiata/internal capsule. This study included 32 subjects with thalamic or putaminal hemorrhage or both. Methods: DTI data were obtained on days 14 to 18. Mean values of fractional anisotropy (FA) within the cerebral peduncle and the corona radiata/internal capsule were analyzed using a computer-automated method. Applying ordinal logistic regression analyses, the ratios between FA values in the affected and unaffected hemisphere (rFA) were modeled in relation to motor outcome scores at 1 month after onset, assessed using the Medical Research Council (MRC) scale (0 = null to 5 = full). Results: For both cerebral peduncle and corona radiata/internal capsule, the relationships between rFA and MRC matched logistic probabilities. While cerebral peduncle rFA values had statistically significant relationships with MRC scores (upper extremity R(2) = 0.271; lower extremity R(2) = 0.191), rFA values for the corona radiata/internal capsule showed less significant relationships (upper extremity R(2) = 0.085; lower extremity R(2) = 0.080). When estimated cerebral peduncle rFA values were <0.7, estimated probability of MRC 0 to 2 was close to 85% for the upper and 60% for the lower extremities. Meanwhile, when estimated rFA values were >0.9, estimated probability for MRC 4 to 5 nearly equaled 50% for the upper and 60% for the lower extremities. Conclusions: FA values from within the cerebral peduncle more accurately predicted motor outcome and is a promising technique for clinical application.
What problem does this paper attempt to address?