(D)
FIGURE 7 Normalized confusion matrices for the best performing models in the age threshold tests 1 (A), 2 (B) and 3 (C). Due to notable class imbalance with the age threshold 3 with the best performing model, the confusion matrix for the best linear SVM model shown additionally (D).
4 | CONCLUSION
The two key findings of the study are 1) fNIRS has potential to classify young and old adults to corresponding age groups with good performance, and 2) the inclusion and exclusion of the subjects in the young, in-between and old age groups affects the classification performance significantly, suggesting different aging rate within these three groups, i.e., young, in-between, and old groups. As the wavelengths used in our fNIRS device are sensitive to hemoglobin and water dynamics, the noticed changes are potentially related to the cerebrovascular and neurohydrodynamic events. However, it is not clear whether water or hemoglobin change caused effects are more significant. Furthermore, the fNIRS was measured from Fp1, and thus the detected changes are only related to the frontal lobe activity and to the hemodynamic and neurohydrodynamic differences in the brain. The effect of superficial layers in fNIRS study of brain aging would be of interest to study more in detail in the future, by utilizing short SD pair in addition to the long SD pair.
Several factors accelerating and decelerating brain aging process have been identified in the literature [2]. The existence of these factors could explain the observed changes in classification performance, when including the subjects in the in-between age groups. Although the study population was considered healthy, the brain aging related accelerative and decelerative factors information, such as alcohol consumption habits and amount of physical activity per week, were not considered in this study. The analysis of these factors would be intriguing, as they may delay or expedite the brain aging process. Furthermore, inclusion of individuals of age over 70 would be of interest as it has been suggested that brain aging accelerates significantly afterwards [36].
Another intriguing prospect is to use fNIRS in the aging study related to sleep. Recently it was found that the brain’s glymphatic system has been found to be more active during the sleep [37] and the aging has been suggested to affect its function negatively [38]. As the system consists of CSF waste clearance, the use of fNIRS to measure the brain pulsations, which are one of the drivers of the glymphatic system [39], and CSF dynamics with water sensitive fNIRS setup could provide potential for gaining further insight into the function of the system in humans, and the effects caused by aging. Furthermore, the EEG of sleep has been successfully utilized in the brain age prediction task. [28] As EEG measured neural activity and fNIRS measured hemodynamic response are connected by neurovascular coupling effect, it could be of interest to study if age-related differences become more pronounced during sleep.
The key limitation of our study is the used relatively small sample size, as in general the use of large datasets increases the statistical power in pattern recognition, [35] and therefore the used datasets in the brain age studies are usually of large size [40]. It would be of great interest in the future to conduct fNIRS brain aging studies with large, well age distributed, sample population to confirm our findings. Additionally, use of multi-channel fNIRS could give boarded picture on the global effects of the aging on brain. Although the used multi-modal measurement setup included different brain measurement modalities, such as magnetic encephalography (MREG) and EEG [21], in this study we focused on the analysis of fNIRS and brain aging specifically due to its novelty in the context. Thus, another area for future exploration in the fNIRS brain aging studies is to analyse combination of brain measurement modalities data, such as fNIRS combined with EEG or MREG. The analysis of different modalities combination would enhance the information content and provide a comparison of the methods performance in the brain aging context. However, although suffering from the lack of data, our study protocol utilized the best practices for ML research with limited data by utilizing nested cross-validation framework, with feature and model selection conducted within the inner loop, for estimating unbiased performance of the models [35]. Additionally, we utilized multiple fNIRS signal features, introduced previously in the study of brain by different measurement modalities. The findings of our study demonstrate the high potential for the use of fNIRS in the study of brain aging.
ACKNOWLEDGMENTS
The authors would like to acknowledge CSC – IT Center for Science, Finland, for providing the computational resources. The study was financed by the Academy of Finland Profi6 funding, 6G-Enabling Sustainable Society (6GESS) programme. Academy of Finland (Terva grant 335723), BF (11146/31/2022), JPND/AoF (357805).
FINANCIAL DISCLOSURE
The authors have no financial conflicts of interest in connection with this article.
CONFLICT OF INTEREST
The authors have no conflicts of interest in connection with this article.
DATA AVAILABILITY STATEMENT
Research data are not shared due to the hospital policy.
REFERENCES
[1] Kontis, V., Bennett, J.E., Mathers, C.D., Li, G., Foreman, K., Ezzati, M., The Lancet , 2017, 389 , 1323.
[2] Turrini, S., Wong, B., Eldaief, M., Press, D.Z., Sinclair, D.A., Koch, G., Avenanti, A., Santarnecchi, E., Ageing Research Reviews , 2023, 88 , 101939.
[3] Higgins-Chen, A.T., Thrush, K.L., Levine, M.E., Seminars in Cell and Developmental Biology , 2021, 116 , 180.
[4] Korhonen, V.O., Myllylä, T.S., Kirillin, M.Y., Popov, A.P., Bykov, A.V., Gorshkov, A.V., Sergeeva, E.A., Kinnunen, M., Kiviniemi, V., IEEE Journal of Selected Topics in Quantum Electronics , 2014,20 , 289.
[5] Ferrari, M., Quaresima, V., NeuroImage , 2012,63 , 921.
[6] Boas, D.A., Gaudette, T., Strangman, G., Cheng, X., Marota, J.J.A., Mandeville, J.B., NeuroImage , 2001, 13 , 76.
[7] Myllylä, T., Harju, M., Korhonen, V., Bykov, A., Kiviniemi, V., Meglinski, I., Journal of Biophotonics , 2018, 11 , e201700123.
[8] Liang, J., Huang, J., Luo, Z., Wu, Y., Zheng, L., Tang, Z., Li, W., Ou, H., Frontiers in Human Neuroscience , 2023, 17 .
[9] Nguyen, T., Kim, M., Gwak, J., Lee, J.J., Choi, K.Y., Lee, K.H., Kim, J.G., Journal of Biophotonics , 2019, 12 .
[10] Boutouyrie, P., Chowienczyk, P., Humphrey, J.D., Mitchell, G.F., Circulation Research , 2021, 128 , 864.
[11] Ferdinando, H., Myllylä, T., in Saratov Fall Meeting 2019: Optical and Nano-Technologies for Biology and Medicine , SPIE, 2020, 11457 , 22.
[12] Mohammadi, H., Peng, K., Kassab, A., Nigam, A., Bherer, L., Lesage, F., Joanette, Y., Neurobiology of Aging , 2021,106 , 103.
[13] Stojan, R., Mack, M., Bock, O., Voelcker-Rehage, C.,NeuroImage , 2023, 273 .
[14] Huang, W., Li, X., Xie, H., Qiao, T., Zheng, Y., Su, L., Tang, Z.-M., Dou, Z., Frontiers in Aging Neuroscience , 2022,14 .
[15] Nóbrega-Sousa, P., Gobbi, L.T.B., Orcioli-Silva, D., Conceição, N.R.D., Beretta, V.S., Vitório, R., Neurorehabilitation and Neural Repair , 2020, 34 , 915.
[16] Lucas, M., Wagshul, M.E., Izzetoglu, M., Holtzer, R.,Journals of Gerontology - Series A Biological Sciences and Medical Sciences , 2019, 74 , 435.
[17] Zeller, J.B.M., Katzorke, A., Müller, L.D., Breunig, J., Haeussinger, F.B., Deckert, J., Warrings, B., Lauer, M., Polak, T., Herrmann, M.J., Brain Imaging and Behavior , 2019, 13 , 283.
[18] Ferdinando, H., Moradi, S., Korhonen, V., Helakari, H., Kiviniemi, V., Myllylä, T., The European Physical Journal Special Topics , 2022, 232 , 655.
[19] Kober, S.E., Spörk, R., Bauernfeind, G., Wood, G.,Neurobiology of Aging , 2019, 81 , 127.
[20] Ilvesmäki, M., Assessment of human brain ageing by functional near-infrared spectroscopy. University of Oulu, 2023.
[21] Korhonen, V., Hiltunen, T., Myllylä, T., Wang, X., Kantola, J., Nikkinen, J., Zang, Y.-F., LeVan, P., Kiviniemi, V., Brain Connectivity , 2014, 4 , 677.
[22] H.s.s, S., Myllylä, T.S., Kirillin, M.Y., Sergeeva, E.A., Myllylä, R.A., Elseoud, A.A., Nikkinen, J., Tervonen, O., Kiviniemi, V.,Quantum Electronics , 2010, 40 , 1067.
[23] Lieslehto, J., Jääskeläinen, E., Kiviniemi, V., Haapea, M., Jones, P.B., Murray, G.K., Veijola, J., Dannlowski, U., Grotegerd, D., Meinert, S., Hahn, T., Ruef, A., Isohanni, M., Falkai, P., Miettunen, J., Dwyer, D.B., Koutsouleris, N., npj Schizophrenia , 2021,7 , 1.
[24] Cole, J.H., Leech, R., Sharp, D.J., Initiative, for the A.D.N.,Annals of Neurology , 2015, 77 , 571.
[25] Löwe, L.C., Gaser, C., Franke, K., Initiative, for the A.D.N.,PLOS ONE , 2016, 11 , e0157514.
[26] Cope, M., The Application of Near Infrared Spectroscopy to Non Invasive Monitoring of Cerebral Oxygenation in the Newborn Infant. University College London, 1991.
[27] Helakari, H., Kananen, J., Huotari, N., Raitamaa, L., Tuovinen, T., Borchardt, V., Rasila, A., Raatikainen, V., Starck, T., Hautaniemi, T., Myllylä, T., Tervonen, O., Rytky, S., Keinänen, T., Korhonen, V., Kiviniemi, V., Ansakorpi, H., NeuroImage: Clinical , 2019,22 , 101763.
[28] Sun, H., Paixao, L., Oliva, J.T., Goparaju, B., Carvalho, D.Z., van Leeuwen, K.G., Akeju, O., Thomas, R.J., Cash, S.S., Bianchi, M.T., Westover, M.B., Neurobiology of Aging , 2019, 74 , 112.
[29] Ferdinando, H., Moradi, S., Korhonen, V., Kiviniemi, V., Myllylä, T., in Neural Imaging and Sensing 2023 , SPIE, 2023,12365 , 18.
[30] Lobier, M., Siebenhühner, F., Palva, S., Palva, J.M.,NeuroImage , 2014, 85 , 853.
[31] Matteo, F., Arjan, H., Phase Transfer Entropy in Matlab. figshare, 2017.
[32] Richman, J.S., Moorman, J.R., American Journal of Physiology-Heart and Circulatory Physiology , 2000, 278 , H2039.
[33] Flood, M.W., Grimm, B., PLOS ONE , 2021, 16 , e0259448.
[34] Borchardt, V., Korhonen, V., Helakari, H., Nedergaard, M., Myllylä, T., Kiviniemi, V., The European Physical Journal Plus , 2021, 136 , 497.
[35] Vabalas, A., Gowen, E., Poliakoff, E., Casson, A.J., PLOS ONE , 2019, 14 , e0224365.
[36] Scahill, R.I., Frost, C., Jenkins, R., Whitwell, J.L., Rossor, M.N., Fox, N.C., Archives of Neurology , 2003, 60 , 989.
[37] Xie, L., Kang, H., Xu, Q., Chen, M.J., Liao, Y., Thiyagarajan, M., O’Donnell, J., Christensen, D.J., Nicholson, C., Iliff, J.J., Takano, T., Deane, R., Nedergaard, M., Science , 2013,342 , 373.
[38] Kress, B.T., Iliff, J.J., Xia, M., Wang, M., Wei, H.S., Zeppenfeld, D., Xie, L., Kang, H., Xu, Q., Liew, J.A., Plog, B.A., Ding, F., Deane, R., Nedergaard, M., Annals of Neurology , 2014,76 , 845.
[39] Bohr, T., Hjorth, P.G., Holst, S.C., Hrabětová, S., Kiviniemi, V., Lilius, T., Lundgaard, I., Mardal, K.-A., Martens, E.A., Mori, Y., Nägerl, U.V., Nicholson, C., Tannenbaum, A., Thomas, J.H., Tithof, J., Benveniste, H., Iliff, J.J., Kelley, D.H., Nedergaard, M.,iScience , 2022, 25 .
[40] Cole, J.H., Franke, K., Trends in Neurosciences , 2017,40 , 681.