Effect Of Thin-section Diffution-weighted Imaging On Acute Cerebral Infarction Diagnosis

Objective： Diffusion-weighted images (DWI) were very sensitive forthe detection of acute cerebral infarctions in currently. Because DW imaginghas limited spatial resolution, still some cases of false-negative diffusion-weighted imaging (DWI) reported. We decreased the section thickness of DWimaging to3mm(thin-section DW imaging)to evaluate the diagnostic value ofthin-section DW imaging on acute cerebral infarction.Materials and Methods: From the consecutive247patients whounderwent conventional MRI studies for a highly suspected cerebral infarctionat our hospital between May2010and December2011, and underwentconventional DWI and thin DWI sequences studies at one time. The time ofonset was longer than two weeks in35patients, diameter of the lesions wasexcess3cm and/or the lesions more than3in32patients, cerebral hemorrhagein4patients, and others reasons (movement artifact, clinical data default,position of lesions was not accord to clinical symptom, et al) excluded15patients, the final161patients were included in the analysis. Two radiologistswho were not aware of the clinical symptoms reviewed all images, cases ofdiscordance between readers were resolves by consensus.All MR examinations were performed by using a1.5-T MR imagingsystem (Signa；GE Medical System). All images were transferred to animaging workstation(Advantage Windows workstation；GE Medical Systems,ADW4.4)and were disposed by a software (Functool, GE Medical Systems),the apparent diffusion coefficient (ADC) and exponential apparent diffusioncoefficient(EADC) maps were created at the same time. The analysis softwareused in this study allowed simultaneous placement of matching Region ofinterests (ROIs) on the DWI, ADC and EADC maps. The lesions were measured which showed on both conventional and thin-section DWIs. Thesignal intensity and size of lesions were measured. A region of interest wascentered in the largest area of diffusion hyperintensity (region of interest area=14~104mm2) and mirrored on to the contralateral hemisphere to obtain aratio of signal intensity(rDWI), and so on rADC, rEADC. The number oflesions on both DWI was recorded. Cases of showing discordance betweenDWIs were resolves by analyzing conventional MRI and follow-up images(within2~5days).It was four outcomes according to both DWIs showing: the lesions wasseen on both DWIs(positive group), the lesions was seen on conventionalDWIs but no thin-section DWIs, the lesions was seen on thin-section DWIsbut no conventional DWIs(thin-section DWIs additional lesions), no lesionswas seen on both DWIs. Based on the vascular distribution area of the lesions,it was divided into anterior and posterior circulation vascular distribution area.It was normal in conventional DWIs, which was defined the false-negativegroup, lesions were showed on the both DWIs, which was defined positivegroup. It was divided into0~24hours and excess24hours groups according tothe interval from onset of symptoms to MRI scan.Statistical analysis: SPSS13.0was used for statistical analysis.Interobserver agreement was determined by using kappa statistics. McNemartest (paired x2test) were compared for the number of lesions betweenconventional and thin-section DWIs, rDWI, rADC, and rEADC werecompared between conventional and thin-section DWIs by using pairedStudent t test. The size, rDWI and rADC between thin-section DWIsadditional lesions and positive groups were compared by using Student t test.Vascular distribution of the thin-section DWIs additional lesions werecompared by using the x2test. The rate of false-negative lesions in vasculardistribution and onset of symptom were conpared by x2test. The statisticalresults (P≤0.05) was considered statistically significant.Results: (1) For the distinction between positive and negative DWIs,the2readersagreed in all of the cases on thin-section DWI,whereas κ=0.895forconventional DWI. After consensus, a total of132patients151lesions onthin-section DWI and146patients166lesions on conventional DWI wasidentified, the number of lesions showing on both DWIs was consideredconspicuous statistically significant (x2=44.406, P=0.001).(2)130patients149lesions were showed on both DWIs,3patients3lesionswere seen on conventional DWIs but not seen on thin-section DWIs,1lesionwas identified magnetic sensitive artifact,2cases were considered cerebralinfarction lesions.16patients17additional lesions were seen on thin-sectionDWIs but not seen on conventional DWIs.1case of on both DWIs wasnormal, which showed1lesion on follow-up DWI.(3) The rDWI of lesions was considered conspicuous statistically significantbetween conventional DWI (rDWIconventional=1.486±0.256) and thin-sectionDWI (rDWIthin=1.617±0.313),(t=-7.574,P=0.000). The rADC value oflesions was considered statistically significant between conventional ADCmap (rADCconventional=0.816±0.170) and thin-section ADC map (rADCthin=0.772±0.188),(t=3.877,P=0.000). The rEADC value of lesions wasconsidered statistically significant between conventional EADC map(rEADCconventional=1.169±0.166) and thin-section EADC map (rEADCthin=1.209±0.172),(t=-3.816,P=0.000).(4) The mean size was considered conspicuous statistically significantbetween thin-section DWI additional lesions (21.235±9.846mm2) and positivegroup lesions (38.283±26.688mm2),(t=5.312,P=0.000). The rDWI wasconsidered conspicuous statistically significant between thin-section DWIadditional lesions (1.485±0.156) and positive group lesions (1.637±0.321),(t=3.296, P=0.002). The rADC was similar between thin-section DWIadditional lesions (0.795±0.160) and positive group lesions (0.772±0.188),(t=-0.479, P=0.632).(5)149cases were finally diagnosed cerebral infarction showed169lesions,67lesions were seen in the anterior and102lesions in the posterior circulation vascular territory.17additional lesions were seen on thin-section DWIs, only1(1.49%) lesion in the anterior circulation vascular territory and16(15.69%)lesions in the posterior circulation vascular territory. The vascular territory ofthin-section DWIs additional lesions was considered conspicuous statisticallysignificant (x2=7.592, P=0.006).(6) A total of169lesions,71lesions occurred within24hours after symptomonset, and98lesions occurred during2~14days after symptom onset.18lesions were missed on conventional DWIs,10lesions occurred within24hours, and8lesions occurred2~14days after symptom onset. False-negativestudies occurred similarly at different time symptom onset (x2=1.215,P=0.270).Conclusion: Thin-section DWI decreases partial volume effect throughreducing the section thickness, and decreases small lesions missing throughrehucing section gap, and increases the sensitivity of DWI for the detection ofsmall ischemic lesions, especially for the lesions in posterior circulationvascular territory. It permitted better conspicuity through increasing NEXand more precise infarction diagnosis.Thin-section DWI examination shouldbe done especially when symptoms persist and are suggestive of this diagnosisbut conventional DWI was showed negative,which can provide a reliableimaging evidence for the early diagnosis and treatment.