This study aims to contribute to empirical and interdisciplinary knowledge on how visual learning analytics tools can support students' cognitive engagement in complex in-class scenarios. Taking a holistic approach, instructional design, learning analytics, and students' perceptions were examined together. The teaching of systematic viewing and image interpretation in radiology education was used to exemplify a complex in-class scenario, and a specific learning dashboard was designed as a support tool. The design was based on both educational and visualization theories and aligned with a pedagogical scenario integrating individual and class-wide activities. The quantity and quality of the cognitive engagement of a group of 26 students were explored. A mixed method approach was used, including computer log file analyses of individual work, analysis of video recordings of in-class small group discussions, and a focus group discussion with the students involved. The in-class scenario with the learning dashboard resulted in a good balance between individual tasks and group discussions and a high degree of active cognitive engagement. Constructive and interactive forms of cognitive engagement were, however, less evident. In addition, the products of these constructive (description of findings) and interactive (type of dialogue) cognitive engagements were of mediocre quality and therefore not optimal for knowledge transfer. The study also showed that the way the students and teacher understood their respective tasks and used the available interaction techniques of the learning dashboard highly influenced the students' cognitive engagement. Finally, several ideas emerged that could help to overcome the deficits found in the training of participants and to improve the tasks set and the learning dashboard itself.