*4.1. Analyzing the Main Features of MuseStudio*

Using as the input a demonstration of MuseStudio, the evaluation of this library was carried out using DESMET [38]. This is a set of techniques applicable to evaluating both software engineering methods and tools. We used the method based on a qualitative case study, which describes a feature-based evaluation. Following the guidelines specified for this technique, an initial list of features that a library or tool for EEG data management should provide was defined (see Table 1). These features were established by two experts (full professors) in cognitive neuropsychology from the University of Castilla-La Mancha (UCLM). As can be observed, some of the features are directly related to the availability of the BIDS.

DESMET was deployed by involving five experts. First, two experts were asked about the main requirements a library for low-cost EEG devices should provide. Second, another three experts were involved to validate MuseStudio by considering the previously proposed requirements. All the experts were professionals with knowledge and skills related to EEG devices, neuroscience, and psychology.

Once Table 1 has been filled in by the experts, DESMET determines the importance degree that should be assigned to each identified feature. Specifically, the importance degrees are Mandatory (M), Highly Desirable (HD), Desirable (D), and Nice to have (N). This importance was also established by the consulted experts.

By using these importance degrees, Table 2 was filled in. As can be noticed, the most important functional and nonfunctional requirements to be supported are signal visualization, import and export data management, and scalability.


**Table 1.** List of features for MuseStudio's evaluation.

**Table 2.** Relevance of features (Mandatory (M), Highly Desirable (HD), Desirable (D), and Nice to have (N)).


Afterwards, according to DESMET, a scale to evaluate each of the described features should be provided. The scale proposed by DESMET (see Table 3) was applied to evaluate each feature according to the following factors: Conformance Acceptability Threshold (CAT) and Conformance score obtained (CSO) for MuseStudio. In particular, three experts (associate professors) from the University of Castilla-La Mancha with experience in the fields emotion recognition, health psychology, and signal processing/computer science agreed about the values of CSOi.


**Table 3.** Judgment scale to assess tool support for a feature.

Once each feature was evaluated, the difference between the CAT and CSO factors was computed as shown in the column Difference (Dif) in Table 4.

Therefore, in order to interpret the values shown in Table 4, the following equations should be considered:

$$Imp\_i = Level\text{ of } relevance\text{ of }each\text{ feature }(i)$$

*CATi* = *Level of support of each feature (i)*

*CSOi* = *Quantitative evaluation of each feature (i) by specialists in several fields*

$$Dif\_i = \mathbb{C}SO\_i - \mathbb{C}AT\_i \tag{1}$$

$$Score\_i = Imp\_i \* Diff\_i \tag{2}$$

$$Total = \sum\_{i=1}^{features} Score\_i \tag{3}$$

We should highlight that a variation of the DESMET method was created. The Importance (Imp) of each feature was weighed using a scale from 1 to 4 (Nice to have—1, Desirable—2, Highly Desirable—3, Mandatory—4). The importance was used to compute the final score of each feature or requirement by multiplying the importance by the difference. This computation is shown in the column Score (Sco) in Table 4. This score is useful for comparing different alternatives, but in our case, the score was only for the MuseStudio's valorization. Lastly, the final score of each technique (Total) was obtained by adding the scores of all the features.

The MuseStudio library achieved a positive total score (15 points). Moreover, it was especially evaluated positively for the "at a distance" feature, since MuseStudio provides full support for exporting the brain activity data. It was also highlighted that the MuseStudio tool has consistency and easily represents the requirements' importance, giving no support to determining which requirements are more important than the others. In any case, MuseStudio provides facilities for data gathering and collection in conformance with the BIDS proposal. Brain data from Muse devices are organized and structured with MuseStudio, and these data can be visualized, imported, exported, and analyzed.


**Table 4.** Results of MuseStudio's evaluation.

In addition, as DESMET suggests, we performed a comparison of the percentage of each feature satisfied by MuseStudio. Figure 6 illustrates the results relative to the considered features. The outcomes of the validation are graphically shown in Figure 6. All previously established requirements were fully achieved. However, additional effort could be made on the user interface feature. At this moment, the information of the sessions and participants must be established directly by modifying this information in different files. Forms may be designed to ease these tasks.

Understanding the score requires knowing how DESMET works. First, the level of importance of a feature was determined by experts without trying the library (between −1 and 5). Thereafter, other experts determined how well implemented a particular feature was (between −1 and 5 again).

The current implementation of MuseStudio satisfies the requirements or features related to visualization, import data, scenario identification, and scalability. Other features of MuseStudio, such as data reviewing and data consistency, are more than satisfied, and the rest are also oversatisfied. At this time, the identified weakness of MuseStudio is that its users need to have certain knowledge about Python, because it does not have a guided user interface yet.

**Figure 6.** Results depending on each feature.
