Keywords
International Medical Informatics Association Yearbook - Sensor informatics - Signal
in-formatics - Imaging informatics - Biomedical informatics - Machine learning - Deep
learning - Personalized medicine
1. Introduction
Sensors, signals, and imaging informatics (SSII) is a vast field of research that
encompasses the acqui-sition, processing, analysis, and interpretation of data. As
our annual reviews often show, the number of publications in this area describing
approaches based on machine or deep learning-based is growing exponentially. This
year, in addition to the growth of publications on DL-based image processing pa-pers,
we observe a decrease in publications on sensors. This can be seen as a reduced interest
in the development of new sensors technologies but can also be interpreted as a reflection
of the replacement of sensor application by sensorless signals and imaging techniques.
Also, the SSII section of the International Medical Informatics Association (IMIA)
Yearbook 2023 presents the survey paper entitled “Alzheimer Disease Detection Studies:
Perspective on Multi-Modal Data” by Dehghani et al. [[1]] This paper explores potential research gaps, challenges, and opportunities related
to automated Alzheimer's disease (AD) detection. The authors provided an overview
of CAD systems for automated AD detection, focusing on different data types, namely,
signals and sensors, medical imaging, and electronic medical records (EMR). As various
medical technologies and com-puter-aided diagnosis (CAD), ranging from biosensors
and raw signals to medical imaging, have been used to provide information about the
state of AD, it may be important for the SSII section to show the challenges and the
remaining gaps to identify research opportunities.
2. Paper Selection Process
2. Paper Selection Process
Due to the expanding field of SSII, we had to change the previously used search terms
and acronyms [[2],[3]] to obtain a manageable number of returns while including the most relevant publications.
We added additional constraints to the previous query and reduced the search terms.
As in the previous years, we limited the search to English-language articles and excluded
all review articles to streamline the review process and executed the queries on Pubmed
and Scopus databases until the end of January 2024. For sensors, signals, and imaging
informatics, this search returned 17, 94, and 326 articles from Pubmed as well as
9, 21, and 194 articles from Scopus, respectively. To focus on the most impactful
journals, we kept the condition that each journal must have at least three publications
in the analyzed subject area to be considered [[2]
[3]
[4]]. After filtering out duplicates, we obtained 12, 32, and 366 (410 in total) publications.
In the next step, we independently graded titles and abstracts on a three-point Likert
scale (1 = not included, 2 = maybe included, 3 = included) and we selected a total
of 58 papers with a cumulative score of eight and above ([Figure 1]). Then, we evaluated the full papers on the same three-point Likert scale and threshold
yielding 14 papers for external review.
Figure 1. Selection process of the best papers for the 2024 IMIA Yearbook of Medical Informatics
for the Sensors, Signals, & Imaging Informatics section.
The total of 14 papers was accessed by eight external reviewers. Based on their ranking,
we selected four papers, all of which were from Imaging Informatics ([Table 1]). However, the clear boundaries between the three fields disappear, as image processing
becomes part of modern sensing (as in the selected paper by Luo et al. [[15]]) and signal informatics (as in the selected paper by Ouzar et al. [[17]]).
Table 1.
Selection result of best papers for the 2024 IMIA Yearbook of Medical Informatics
for the Section Sensors, Signals, and Imaging Informatics. The articles are listed
in alphabetical order of the first author's surname. A content summary of these best
papers can be found in the appendix of this synopsis.
|
Section Sensors, Signals, and Imaging Informatics
|
|
Li Z, Fan Q, Bilgic B, Wang G, Wu W, Polimeni JR, Miller KL, Huang SY, Tian Q. Diffusion
MRI data analysis assisted by deep learning synthesized anatomical images (DeepAnat).
Med Image Anal. 2023 May;86:102744. doi: 10.1016/j.media.2023.102744.
Chen Y, Lu X, Xie Q. Collaborative networks of transformers and convolutional neural
networks are powerful and versatile learners for accurate 3D medical image segmentation.
Comput Biol Med. 2023 Sep;164:107228. doi: 10.1016/j.compbiomed.2023.107228.
Luo M, Yang X, Wang H, Dou H, Hu X, Huang Y, Ravikumar N, Xu S, Zhang Y, Xiong Y,
Xue W, Frangi AF, Ni D, Sun L. RecON: Online learning for sensorless freehand 3D ultrasound
reconstruction. Med Image Anal. 2023 Jul;87:102810.
Ouzar Y, Djeldjli D, Bousefsaf F, Maaoui C. X-iPPGNet: A novel one stage deep learning
architecture based on depthwise separable convolutions for video-based pulse rate
estimation. Comput Biol Med. 2023 Mar;154:106592.
|
3. Outlook
Fourteen papers were pre-selected as candidate best papers for 2023, three of them
from the field of sensor informatics, three from signal informatics, and eight from
imaging informatics.
Of the pre-selected papers from the field of sensor informatics, the first introduces
a cuffless blood pressure monitor that does not require periodic calibration. In this
study, Wang et al. [[5]] showed that the blood pressure monitor generates high-quality photoplethysmographic
signals with satisfactory accuracy both at initial calibration and 1-month follow-ups.
The paper by Kumaki et al. [[6]] was also pre-selected as a best paper candidate, presenting a sheet-type sensor
for heartbeat monitoring of sleeping infants and young children, while the third candidate
paper by Ortega-Rodríguez et al. [[7]] presents a method for reducing the number of electrodes in EEG-based biometric
systems for identify-ing individuals by EEG.
As for the field of signal informatics, the pre-selected papers ranged from Covid-19
detection by Celik [[8]] based on cough, breath and voice information, emotion recognition by EEG by Qiu
et al. [[9]], and intelligent antepartum fetal monitoring by Cao et al. [[10]]. In the end, all three papers use a DL frame-work to accomplish the proposed task.
Finally, we can categorise the eight papers pre-selected from the field of imaging
informatics into three main groups: image segmentation [[11]
[12]
[13]
[14]], image reconstruction [[15],[16]] and image analysis [[17]]. As in the field of signal informatics, all the aforementioned pre-selected papers
on image processing also propose DL-based methods to perform the desired task. The
only pre-selected paper in the field of imaging informatics that does not propose
a DL-based solution is the one published by Nie et al. that presents a dataset of
oral implant image [[18]].
4. Discussion
While utmost care was placed on the creation of the search queries, the sheer number
of publications in SSII made it necessary to limit the scope to achieve a manageable
amount or results. This emphasizes the need for continuous refining of the search
queries based on emerging technologies, terms, and applications.
Deep learning-based approaches for medical image and biosignal processing dominate
the selected papers for this year. Most of these papers deal with medical image processing.
Notably, among the top four papers, the first sensorless concepts based on deep learning-based
image processing were present-ed. These approaches attempt to replace biosignal information,
typically measured by biomedical sen-sors, with image information. Sensorless biosignal
processing using ML/DL-based methods thus rep-resents an intriguing and pioneering
development in this field.
For the 2025 Yearbook, we plan to rename our section from “Sensors, Signals, and Imaging
Informat-ics (SSII)” to “Biosignal and Imaging Informatics (BII)” to better reflect
the two SSII subsections—sensors and signals—by the term “biosignal.” The main focus
is not on newly developed sensor tech-nology, but on the methods for processing and
analyzing the bio-information generated by the sensors, which motivated us to rename
the SSII section.
To summarise, the renaming of the section to “Biosignal and Imaging Informatics (BII)”
is a strategic move that is in line with the current trends and future direction of
the field. This change emphasises the shift from sensor development more to advanced
processing and analysis of biosignals and medical images, which ultimately contributes
to better patient outcomes and more efficient healthcare delivery.
