KOSMOS – Measurements

Competitive Analysis

Understand the market and gain background knowledge to support the future work.

With the above discussed directions, my first step was to clarify the project scope. With the PM's help, I had opportunities to talk with several cardiologists and cardiac sonographers in our network to collect a couple of basic measurements to start with.

By reading some echocardiography textbooks and articles, I was able to gain basic knowledge in cardiac measurements to support myself and the team's future work. Through reading the user guides of other ultrasound products in the market, watching online cardiac scanning videos, and operating other competitor's products, I was able to learn which measurements in the market were supported, and how user flow and interactions for cardiac measurements could be designed. By presenting the summary of what I learned from the research, I was able to help the team gain sufficient background knowledge to move forward with me.

Ideation

Brainstorm ideas with other design team members.

With some design ideas in mind, I started ideating potential user flow and interactions. In this stage, I also invited the other two designers on the team to join the ideation sessions. Since we were all working from home during the period, we utilized InVision Freehand to present and discuss our ideas.

User Flow & User Need Collection

Collect user needs with a user flow diagram and propose potential solutions accordingly.

Through multiple rounds of discussions between designers and the PM, I was able to draw a user flow diagram with more details based on our assumptions.

In this early stage, I explored two main user flow approaches: free flow, and protocol-driven flow. In the free flow, users could acquire images/clips first, and chose the measurements needed later. They could conduct the measurements instantly in the cine review or afterward in the exam review. They could also convert a regular measurement into a particular measurement.

While in the protocol-driven flow, we supposed that users already knew what they wanted to conduct in the very beginning, so they would choose a measurement type first before starting acquiring images/clips. This idea was inspired by what we already had for Ejection Fraction measurements in the system.

Next, we thought it could be a good timing to start collecting feedback from our users, so I created mid-fi wireframes combining with the user flow diagram to present to cardiac sonographers in our network. In this diagram, more complex scenarios such as conducting multiple measurements across different views and clips to generate calculations were involved.

From the user feedback, I got four main takeaways:

	A particular measurement would not be used independently. Users would use a whole calculation package instead, so we should provide measurement packages instead of independent measurements for users to choose.
	The free flow had higher flexibility which made more sense to users because they might need to conduct a set of measurements during the scan, but they might not know it before starting an exam. Therefore, we would get rid of the protocol-driven flow and go with the free flow option.
	Converting a regular measurement into a particular measurement was not a common use case and it would take effort to manage and implement, so we could de-prioritize this functionality.
	For some measurement packages, measurements would be conducted within different clips and different modes, so this complex scenario might need to be covered in the future as well.

Besides, using AI to support view selection and caliper placing would be very helpful for users, but after talking with the AI team, we realized that it would took time and effort to implement. Therefore, for the first release, everything would be manual, but AI-assistant measurements could be a direction to work on in the future.

Through multiple rounds of discussion, I was also able to clarify the requirements for the new features and listed down potential solutions accordingly.

Design Decisions

Support easy navigation, operation, and info access.

After gathering initial feedback from users and having the project scope defined, I made several design decisions to support intuitive and smooth operation:

All the calculations that could be calculated based on the existing measurements would be automatically generated and presented to support users efficiently gain their desired information.

Reference images which found to be helpful for users would be provided to effectively guide users with clarity during the process.

Visual feedback for the measurement operation progress would be designed to support easy navigation, such as marking the progress on the control panel and the cine bar.

To help users focusing and avoid the crowded and distracting views, clips with basic measurements and annotations would not be allowed to add particular measurements, vice versa.

Only saving a clip instead of saving an image would be allowed in conducting particular measurements, in order to support the scenario of multiple measurements of a package placed in different frames of a clip, and to simplify the workflow at the same time.

Since we plan to have AI-assistant measurements in the future, we would keep this in mind and consider scalability while working on the design.

Project Scope

Define the measurements and calculations to support initially based on the market needs.

In addition to the design direction, I also clarify the project scope parallelly. Through the above mentioned research, I found over 200 cardiac measurements and calculations across different scan modes. To define which measurements to give priorities to, I reached out to over 10 physicians and cardiac sonographers.

By continually discussing with the PM and sonography professionals and through multiple iterations, I eventually narrowed down the scope to the most frequently used ones (around 20) and generated the list to support. The measurement packages in the B mode (the brightness scan mode) were given higher priorities which would be implemented in the first release.

Design Wireframes

Utilize wireframes with higher fidelity to collect further interaction and visual feedback.

With the scope defined, based on the previous user feedback and the design decisions, I created mid-to-high fidelity wireframes and utilized them to collect further feedback from sonography professionals. Over a course of few months, more than 10 cardiac sonographers got involved, who also talked with our sales and direct customers in a regular basis. The way we interacted was more like a focus group combining with participatory design method. Since sonographers with different training backgrounds had various opinions on how to categorize measurements, I applied this method for them to give me feedback and discuss different opinions sufficiently with the team so that we could reach to a solid conclusion efficiently. In the process, developers were also involved in an early stage.

Participatory Design

Improve the interaction and visual design to support efficient navigation and operation of the product based on the feedback of sonography professionals.

Through several rounds of feedback collection sessions, discussions, and iterations, I refined and polished the design accordingly. Below I took the measurement package with the most crowded view – "P L/SAX (Parasternal Long-axis and Short-axis)" package, as an example to demonstrate what I learned and the main changes made along the way:

Users might need to conduct a set of measurements during a scan section and different users might have their different preferred order for conducting the measurements. Therefore, a predefined order was provided following the norm in the field, but users could change the order by tapping on a particular measurement caliper trigger, or skipping the current one and coming back later. I also found that users would choose a qualified view first, and placed all the calipers belonging to one particular view in the same view, so I modified the user flow by giving a view setting control for users to pick a view before starting choosing a measurement caliper. Besides, I also learned that users did not need to see the measurement values and its caliper triggers in a view which they did not belong to, so it was addressed in the design to make each view cleaner as well.

Users needed to know the current status and the next steps in the procedure, so on the control panel, a menu of all the measurements in the package for the current view was designed with the active measurement label highlighted. To enhance the navigation, the visual design and the layout for the controls were improved as well. The view switching control was changed from tapping to switch to a dropdown selection, which made the control looks more like a button and be more efficient when there were more than two views in one package for users to choose. Besides, a visual representative of a particular view was added on to the set view button to gave user a clear connection between the button on the control panel and the control on the cine bar of the imaging window. I also improved the visual treatment for the uncompleted status on the menu by changing a solid line with a tint color of the theme color into a dotted line with the theme color, to make it easier to perceive.

From user feedback, I realized that colors were helpful for differentiating different types of measurements. Therefore, different tertiary colors in the system were leveraged to code different views, such as ED (end diastole), ES (end systole), MS (mid systole), and others. The controls and the calipers that belonged to the particular views were all color coded.

Cine bar was a control for users to review an acquired clip. Users could use the left and right fences to pick a desired range to save. For the package with measurements required in diastolic and systolic views, users would conduct them in one cardiac cycle. If users acquired a clip which was over a certain length, the two views would be super close on the cine bar which made it hard to operate the set view controls. Also, because of the current cine bar design was not clean enough, we took it as an opportunity to improve the cine bar by refining the visual treatments and adding a zoom feature. We learned that the speed control and the next/previous controls were not used frequently in the scenario of conducting measurements, so we decided to remove them from the cine bars for measurements. Eventually, we figured out a new set of cine bars – one for the regular cine review / exam review, the other for the clips with measurements. For the later one, users could adjust the begin and the end fences to select a range to zoom in. By tapping the "..." icon button or the legend on the right side, they could return to the initial view.

Users needed a place to check out all the conducted measurements and all the corresponding calculations. Also, after acquiring measurements and leaving the cine review, users might want to adjust the calipers. Therefore, a cardiac report with all the measurement and calculation values auto-generated was designed. Users could edit the calipers in the exam review edit view. They were also allowed to conduct one package multiple times in one scanning exam.

Final Design

Demonstrate a complete user flow of the measurement and calculation new feature with the scenario of conducting a measurement package in different views of a clip.

At the end of the project, the wireframes were uploaded to Zeplin and handed over to developers with Jira tickets. The component library for the product was also updated along the way.

The final design was presented as below:

After acquiring a desired view, users could go to the cine review and choose a desired measurement package.

Later, users would pick a view on a cine bar, and started measuring. They could choose whichever preferred order and skipped the ones they did not need.

The help feature was presented as pop-up slides for user reference. They could check out and focus on the guidance that belonged to the current package.

When users were done with the measuring work in the first step (in ED), they could switch to the next step (in ES). By zooming in on the cine bar, they could better operate the view setting control for the next step.

If users were not satisfied with what they had got, they could remove the current package by tapping the "remove all" button.

In the end, users could save the clip along with the current measurement package, and checked it out or edited it in exam review and report.

Hi-fi Prototypes

Impacts

Facilitate data-supported cardiac diagnosis.

Based on the feedback from cardiac sonographers, the cardiac measurements and calculation features allowed them to conduct measurements more accurately and gain calculation values efficiently. With the numbers, they were able to quickly tell the issues from different ranges. One example was the measurements for the diameter of aorta root (AoR) and left atrium (LA). With the calculation LA/Ao value, physicians could estimate the increase in effective pulmonary blood flow since the ratio correlates significantly with increased pulmonary flow attributable to excessive transductal flow.

Because of this project, we also built a closer relationship with sonographers in our network. The cooperation model of involving sonography professionals in early design stages and collecting feedback along the way could be applied to other projects in the future.

Reflection

Limitation

	Only those core measurements that could be conducted in B mode were covered.
	Without AI's help, some measurements may not be conducted so accurately, such as those required in the mid-systolic view.

Next Step

	Collect feedback from usability testings and the market.
	Work on the measurements belonging to the core cardiac measurement scope but conducted in other ultrasound modes.
	Develop AI-assistant features for measurements to enhance accuracy and avoid mistakes.

What I Learned

	Effectively communicating with different roles in a project is crucial to keeping the team members on the same page and moving forward with us, especially when the scope and the direction are ambiguous.
	Different user groups may have different needs, and it may be hard to please all the users. Without customization, what we could do is trying our best to cover all the contents and functionalities that target users need, but avoid adding operation burden to cater smooth user experience in the design.

	Define the scope of the cardiac measurement and calculation lists that shall be prioritized based on target user needs and to be built initially in the first release.
	Deisgn a smooth user flow for measuring, reviewing, editing and exporting based on the existing flow.
	Allow users to conduct measurements in a package with their preferred order.
	Support users to conduct measurements in different frames of a clip, and even across different clips and ultrasound modes, either instantly or later on.
	Present measurement calipers and values, and the calculation values in an efficient way across different stages of the user flow.