CAJ | 학술논문

目的探讨超声回波强度界面多级分解法定量分析离体猪横纹肌微波消融热损伤、无水乙酸注射化学损伤后的超声图像纹理变化的应用价值。方法分别以微波消融、无水乙酸注射损伤离体猪横纹肌作为实验研究模型，采用计算机纹理特征自动提取定量分析--超声回波强度界面多级分解法，针对不同损伤区域的横纹肌声像图纹理变化进行对照研究，判断各组所提取的8种反映肌纤维纹理分布数学几何特征参量的统计学差异。结果与正常区域（ROI1）比较，微波消融损伤区域（ROI2）用以定量描述纹理灰度的均值（Mean）、纹理基元形状的不规则度（IRGL）、纹理分布的周期性（POD）的3种纹理特征差异有统计学意义（Mean：1.9143±0.2914 vs 1.2334±0.3357，t=-5.306， P=0.000；IRGL：0.5577±0.0334 vs 0.5092±0.0459，t=-2.957，P=0.007；POD：0.00027±0.000005 vs 0.00029±0.000008，t=4.782，P=0.000）；与正常区域（ROI1）比较，无水乙酸注射损伤区域（ROI3）用以定量描述纹理的密致性（NOB）、纹理基元的平均大小（SOB）、POD的3种纹理特征差异有统计学意义（NOB：51.0324±13.6998 vs 31.6042±4.8315，t=4.633，P=0.000；SOB：16.4843±3.9349 vs 25.6230±2.3555，t=6.903，P=0.000；POD：0.00026±0.000015 vs 0.00029±0.000008， t=-4.459， P=0.000）；两组理化损伤区域相比，Mean、IRGL、NOB、SOB的4种纹理特征差异有统计学意义（Mean：1.2664±0.2688 vs 1.9143±0.2914， t=-5.661，P=0.000；IRGL：0.5220±0.0422 vs 0.5577±0.0334， t=-2.295，P=0.032；NOB：51.0324±13.6998 vs 34.5856±2.6362，t=4.048，P=0.000；SOB：16.4843±3.9349 vs 25.3176±2.3501，t=-6.676，P=0.000）。结论超声回波强度界面多级分解法能够有效提取横纹肌超声图像中的肌纹理形态、组构及分布状态信息，定量反映出不同理化损伤类型导致横纹肌纹理声像图差异的显著性，初步具备了定量诊断能力。目的探討超聲迴波彊度界麵多級分解法定量分析離體豬橫紋肌微波消融熱損傷、無水乙痠註射化學損傷後的超聲圖像紋理變化的應用價值。方法分彆以微波消融、無水乙痠註射損傷離體豬橫紋肌作為實驗研究模型，採用計算機紋理特徵自動提取定量分析--超聲迴波彊度界麵多級分解法，針對不同損傷區域的橫紋肌聲像圖紋理變化進行對照研究，判斷各組所提取的8種反映肌纖維紋理分佈數學幾何特徵參量的統計學差異。結果與正常區域（ROI1）比較，微波消融損傷區域（ROI2）用以定量描述紋理灰度的均值（Mean）、紋理基元形狀的不規則度（IRGL）、紋理分佈的週期性（POD）的3種紋理特徵差異有統計學意義（Mean：1.9143±0.2914 vs 1.2334±0.3357，t=-5.306， P=0.000；IRGL：0.5577±0.0334 vs 0.5092±0.0459，t=-2.957，P=0.007；POD：0.00027±0.000005 vs 0.00029±0.000008，t=4.782，P=0.000）；與正常區域（ROI1）比較，無水乙痠註射損傷區域（ROI3）用以定量描述紋理的密緻性（NOB）、紋理基元的平均大小（SOB）、POD的3種紋理特徵差異有統計學意義（NOB：51.0324±13.6998 vs 31.6042±4.8315，t=4.633，P=0.000；SOB：16.4843±3.9349 vs 25.6230±2.3555，t=6.903，P=0.000；POD：0.00026±0.000015 vs 0.00029±0.000008， t=-4.459， P=0.000）；兩組理化損傷區域相比，Mean、IRGL、NOB、SOB的4種紋理特徵差異有統計學意義（Mean：1.2664±0.2688 vs 1.9143±0.2914， t=-5.661，P=0.000；IRGL：0.5220±0.0422 vs 0.5577±0.0334， t=-2.295，P=0.032；NOB：51.0324±13.6998 vs 34.5856±2.6362，t=4.048，P=0.000；SOB：16.4843±3.9349 vs 25.3176±2.3501，t=-6.676，P=0.000）。結論超聲迴波彊度界麵多級分解法能夠有效提取橫紋肌超聲圖像中的肌紋理形態、組構及分佈狀態信息，定量反映齣不同理化損傷類型導緻橫紋肌紋理聲像圖差異的顯著性，初步具備瞭定量診斷能力。
목적탐토초성회파강도계면다급분해법정량분석리체저횡문기미파소융열손상、무수을산주사화학손상후적초성도상문리변화적응용개치。방법분별이미파소융、무수을산주사손상리체저횡문기작위실험연구모형，채용계산궤문리특정자동제취정량분석--초성회파강도계면다급분해법，침대불동손상구역적횡문기성상도문리변화진행대조연구，판단각조소제취적8충반영기섬유문리분포수학궤하특정삼량적통계학차이。결과여정상구역（ROI1）비교，미파소융손상구역（ROI2）용이정량묘술문리회도적균치（Mean）、문리기원형상적불규칙도（IRGL）、문리분포적주기성（POD）적3충문리특정차이유통계학의의（Mean：1.9143±0.2914 vs 1.2334±0.3357，t=-5.306， P=0.000；IRGL：0.5577±0.0334 vs 0.5092±0.0459，t=-2.957，P=0.007；POD：0.00027±0.000005 vs 0.00029±0.000008，t=4.782，P=0.000）；여정상구역（ROI1）비교，무수을산주사손상구역（ROI3）용이정량묘술문리적밀치성（NOB）、문리기원적평균대소（SOB）、POD적3충문리특정차이유통계학의의（NOB：51.0324±13.6998 vs 31.6042±4.8315，t=4.633，P=0.000；SOB：16.4843±3.9349 vs 25.6230±2.3555，t=6.903，P=0.000；POD：0.00026±0.000015 vs 0.00029±0.000008， t=-4.459， P=0.000）；량조이화손상구역상비，Mean、IRGL、NOB、SOB적4충문리특정차이유통계학의의（Mean：1.2664±0.2688 vs 1.9143±0.2914， t=-5.661，P=0.000；IRGL：0.5220±0.0422 vs 0.5577±0.0334， t=-2.295，P=0.032；NOB：51.0324±13.6998 vs 34.5856±2.6362，t=4.048，P=0.000；SOB：16.4843±3.9349 vs 25.3176±2.3501，t=-6.676，P=0.000）。결론초성회파강도계면다급분해법능구유효제취횡문기초성도상중적기문리형태、조구급분포상태신식，정량반영출불동이화손상류형도치횡문기문리성상도차이적현저성，초보구비료정량진단능력。
Objective To quantitatively analyze and compare the texture features of thermal and chemical lesions on the porcine striated muscle, in vitro extracted from high-frequency ultrasonograms using computer-assisted image analysis technique, and to investigate the application values. Methods The thermal lesion and chemical lesion were induced in vitro in porcine striated muscle by microwave ablation and anhydrous acetic acid injection, respectively. The two dimension (2D) ultrasonographic ifndings were qualitatively compared between the groups of thermal and chemical lesion models, in which eight textural features in geometric mathematics extracted from 2D ultrasonograms were quantitatively analyzed by a technique of computer-assisted image analysis named multiscale decomposition method of echo intensity of interface relfections. Results As expected, microwave ablation and anhydrous acetic acid caused signiifcant changes of several texture features extracted from ultrasonograms. There were significant differences between the normal group and microwave ablation group in grayscale mean (Mean), irregularity (IRGL) and periodicity of distribution (POD) as follows (Mean: 1.9143±0.2914 vs 1.2334±0.3357, t=-5.306, P=0.000; IRGL: 0.5577±0.0334 vs 0.5092±0.0459, t=-2.957, P=0.007; POD: 0.000 27±0.000 005 vs 0.000 29±0.000 008, t=4.782, P=0.000). There were signiifcant differences between the normal group and anhydrous acetic acid injection group in number of blobs (NOB), size of blobs (SOB) and periodicity of distribution (POD) as follows (NOB: 51.0324±13.6998 vs 31.6042±4.8315, t=4.633, P=0.000; SOB:16.4843±3.9349 vs 25.6230±2.3555, t=6.903, P=0.000;POD:0.000 26±0.000 015 vs 0.000 29±0.000 008, t=-4.459, P=0.000). For each group of injured regions, there were significant differences between the microwave ablation group and anhydrous acetic acid injection group in Mean, IRGL, NOB and SOB as follows (Mean: 1.2664±0.2688 vs 1.9143±0.2914, t=-5.661, P=0.000; IRGL: 0.5220±0.0422 vs 0.5577±0.0334, t=-2.295, P=0.032;NOB:51.0324±13.6998 vs 34.5856±2.6362, t=4.048, P=0.000;SOB:16.4843±3.9349 vs 25.3176±2.3501, t=-6.676, P=0.000). Conclusion Technique of computer-assisted image analysis named multiscale decomposition method of echo intensity of interface relfections, based on multiscale blob features extraction, was useful to differentiate ultrasonic texture features between the groups injured in our study, which established quantitative muscle ultrasound as a practical and reliable tool for the muscle injury diagnosis to distinguish the structural changes induced by different physiochemical factors.