Faculty of Computer Science, Otto-von-Guericke University Magdeburg, Universitätsplatz 2, D-39106, Magdeburg, Germany.
University Hospital Magdeburg, Germany, Leipziger Str. 44, D-39120, Magdeburg, Germany.
Int J Comput Assist Radiol Surg. 2020 Sep;15(9):1525-1535. doi: 10.1007/s11548-020-02217-9. Epub 2020 Jul 4.
Medical case-based reasoning solves problems by applying experience gained from the outcome of previous treatments of the same kind. Particularly for complex treatment decisions, for example, incidentally found intracranial aneurysms (IAs), it can support the medical expert. IAs bear the risk of rupture and may lead to subarachnoidal hemorrhages. Treatment needs to be considered carefully, since it may entail unnecessary complications for IAs with low rupture risk. With a rupture risk prediction based on previous cases, the treatment decision can be supported.
We present an interactive visual exploration tool for the case-based reasoning of IAs. In presence of a new aneurysm of interest, our application provides visual analytics techniques to identify the most similar cases with respect to morphology. The clinical expert can obtain the treatment, including the treatment outcome, for these cases and transfer it to the aneurysm of interest. Our application comprises a heatmap visualization, an adapted scatterplot matrix and fully or partially directed graphs with a circle- or force-directed layout to guide the interactive selection process. To fit the demands of clinical applications, we further integrated an interactive identification of outlier cases as well as an interactive attribute selection for the similarity calculation. A questionnaire evaluation with six trained physicians was used.
Our application allows for case-based reasoning of IAs based on a reference data set. Three classifiers summarize the rupture state of the most similar cases. Medical experts positively evaluated the application.
Our case-based reasoning application combined with visual analytic techniques allows for representation of similar IAs to support the clinician. The graphical representation was rated very useful and provides visual information of the similarity of the k most similar cases.
基于病例的推理通过应用从以前相同类型的治疗结果中获得的经验来解决问题。特别是对于复杂的治疗决策，例如偶然发现的颅内动脉瘤 (IA)，它可以为医学专家提供支持。IA 存在破裂的风险，可能导致蛛网膜下腔出血。需要仔细考虑治疗，因为对于破裂风险低的 IA，可能会带来不必要的并发症。通过基于以前病例的破裂风险预测，可以支持治疗决策。
我们提出了一种用于 IA 基于病例推理的交互式可视化探索工具。在有新的感兴趣的动脉瘤的情况下，我们的应用程序提供了可视化分析技术，以识别在形态方面最相似的病例。临床专家可以获得这些病例的治疗，包括治疗结果，并将其转移到感兴趣的动脉瘤。我们的应用程序包括热图可视化、自适应散点图矩阵以及完全或部分有向图，带有圆形或力导向布局，以指导交互式选择过程。为了满足临床应用的需求，我们进一步集成了异常病例的交互式识别以及用于相似性计算的交互式属性选择。我们使用了一份由六名经过培训的医生组成的问卷评估。
我们的应用程序允许基于参考数据集进行 IA 的基于病例的推理。三个分类器总结了最相似病例的破裂状态。医学专家对该应用程序给予了积极评价。
我们的基于病例的推理应用程序结合了可视化分析技术，允许表示相似的 IA 以支持临床医生。图形表示被评为非常有用，并提供了 k 个最相似病例相似性的可视化信息。
