Siiri Rautio

I am a postdoctoral researcher working in the inverse problems research group, at the mathematics department of University of Helsinki, Finland.

My main interest in research is combining deep learning with rigorous analytical mathematical methods, developing algorithms for inverse problems that take advantage of both approaches. Currently, many deep learning methods can be described as a ‘black box’, which reduces their credibility from the viewpoint of an end user, for example a doctor performing medical diagnosis. By keeping the machine-learned parts of complicated methods fairly limited, the results are more explainable and trustworthy.

I am mainly interested in medical imaging applications. I have experience in limited-angle and sparse X-ray tomography (2D, 3D), for example digital breast tomosynthesis and endodontic CBCT imaging, and also electrical impedance tomography. I regularly work with experimental and patient data.

As a scientist, I am motivated by my work having a tangible, positive impact on the world.

ResearchGate

Publications:

Juan Pablo Agnelli, Fernando Moura, Siiri Rautio, Melody Alsaker, Rashmi Murthy, Matti Lassas, Samuli Siltanen. Stroke classification using virtual hybrid edge detection from in silico electrical impedance tomography data. Inverse Problems and Imaging. doi: 10.3934/ipi.2025059.
Sara Sippola, Siiri Rautio, Andreas Hauptmann, Takanori Ide, Samuli Siltanen. Learned enclosure method for experimental EIT data. Applied Mathematics for Modern Challenges, 2025, 4: 47-65.
Siiri Rautio, Ill-posed Microlocal Imaging Using Targeted Deep Learning. Doctoral dissertation, 2025.
Melody Alsaker, Siiri Rautio, Fernando Moura, Juan Pablo Agnelli, Rashmi Murthy, Matti Lassas, Jennifer L. Mueller, Samuli Siltanen. CT scans without X-rays: parallel-beam imaging from nonlinear current flows. Applied Mathematics for Modern Challenges, 2025, 5: 1-35.
Siiri Rautio, Rashmi Murthy, Tatiana A Bubba, Matti Lassas, Samuli Siltanen. Learning a microlocal prior for limited-angle tomography. IMA Journal of Applied Mathematics, 2023, 88.6: 888-916.
Tatiana Bubba, Luca Calatroni, Ambra Catozzi, Serena Crisci, Thomas Pock, Monica Pragliola, Siiri Rautio, Danilo Riccio, Andrea Sebastiani. Bilevel learning of regularization models and their discretization for image deblurring and super-resolution. Advanced Techniques in Optimization for Machine Learning and Imaging. ATOMI 2022. Springer INdAM Series, vol 61. Springer, Singapore.
Siiri Rautio, Combining Deep Learning and Iterative Reconstruction in Dental Cone-Beam Computed Tomography. Master's thesis in collaboration with Planmeca Oy, 2019.

Positions of responsibility:

Coordinator of FCAI and FAME Special interest group: AI for inverse problems and imaging. (2024 - present)
Secretary and board member of the Finnish Inverse Problems Society. (2020 - 2022)

Conference presentations and invited seminar talks:

AI Day 2025 (November 13, 2025 in Espoo, Finland) Learned enclosure method for experimental EIT data (talk)
International Conference on IMAGING25 (September 22–25, 2025 in Varenna, Italy) Metal segmentation and reconstruction of CBCT sinograms with 3D complex wavelets (talk)
Applied Inverse Problems 2025 (July 28 to August 1, 2025 in Rio de Janeiro, Brazil) Real-data EIT reconstruction using virtual X-rays and deep learning (talk)
Sustainable Tomographic Imaging with Learning and Regularization (June 16, 2025 in Bologna, Italy) Computed tomography without X-rays: parallel-beam imaging from nonlinear current flows (invited talk)
Inverse Problem Seminar from Theory to Applications -Finland-Japan Seminar for Inverse Problem- (January 27-28, 2025 in Tokyo, Japan) Computed tomography without X-rays: parallel-beam imaging from nonlinear current flows (talk)
Ehime University (January 24, 2025 in Matsuyama, Japan) Electrical impedance tomography and virtual X-rays (invited seminar talk)
RIMS Symposium: Theoretical and numerical approaches for inverse problems (January 20-22, 2025 in Kyoto, Japan) Computed tomography without X-rays: parallel-beam imaging from nonlinear current flows (invited talk)
30th Inverse Days (December 10-13, 2024, in Oulu, Finland) CT scans without X-rays: parallel-beam imaging from nonlinear current flows (talk)
Colorado State University (October 17, 2024 in Fort Collins, Colorado, US) Learning to deblur virtual hybrid parallel-beam sinograms (seminar talk)
MATRIX × IMAGINARY (August 31-September 4, 2024 in Les Diablerets, Switzerland), Hands-on tomography maths (workshop)
Finland-Japan Workshop in Industrial and Applied Mathematics (August 26-30, 2024 in Helsinki, Finland), CT scans without X-rays: parallel-beam imaging from nonlinear current flows (talk)
33rd European Conference on Operational Research (June 30-July 3, 2024 in Copenhagen, Denmark), Real data EIT reconstruction using virtual X-rays and deep learning (talk)
SIAM Conference on Imaging Science (May 28-31, 2024 in Atlanta, USA), Real data EIT reconstruction using virtual X-rays and deep learning (talk)
FinTomo Seminar 2024 (April 18-19, 2024 in Jyväskylä, Finland), Real data EIT reconstruction using virtual X-rays and deep learning (talk)
94th Annual meeting of the International Association of Applied Mathematics and Mechanics (March 18-22, 2024 in Magdeburg, Germany), Real data EIT reconstruction using virtual X-rays and deep learning (talk)
29th Inverse Days (December 11-15, 2023 in Lahti, Finland), Real data EIT reconstruction using virtual X-rays and deep learning (talk)
AIX Forum: Inverse problems (November 29th, 2023 in Helsinki, Finland), Reformulating electrical impedance tomography as virtual X-ray tomography with the help of deep learning (talk)
AI Day 2023 (November 8th, 2023 in Espoo, Finland), Electrical impedance reconstruction using virtual X-rays and machine learning (poster)
11th Applied Inverse Problems (September 4-8, 2023 in Göttingen, Germany), Electrical impedance reconstruction using virtual X-rays and machine learning (poster)
10th International Congress on Industrial and Applied Mathematics (August 20-25, 2023 in Tokyo, Japan), New insight on electrical impedance reconstruction using virtual X-rays and machine learning (talk)
Mathematics of electrical imaging: modeling, theory and implementation (June 12-14 in Toulouse, France), Electrical impedance reconstruction using virtual X-rays and machine learning (talk)
Rich and non-linear tomography in medical imaging, materials and non destructive testing (March 27-31, 2023 in Cambridge, UK), Electrical impedance reconstruction using virtual X-rays and machine learning (talk)
28th Inverse Days (December 12-16, 2022 in Kuopio, Finland), New Insight into EIT reconstruction using virtual X-rays and machine learning (talk)
Advanced Techniques in Optimization for Machine learning and Imaging (ATOMI) workshop (June 20-24, 2022 in Rome, Italy), Learning a microlocal prior for limited-angle tomography (pitch talk)
2nd HMI Lanczos Workshop on Continuous and discrete iterative methods for image and signal reconstruction (April 26-30, 2022 in Dublin, Ireland), Learning a microlocal prior for limited-angle tomography (poster)
Mathematics for deep learning opening workshop (April 21-22, 2022 in Bath, UK), Learning a microlocal prior for limited-angle tomography (poster)
SIAM Conference on Imaging Science (March 21-25, 2022 virtual), Learning a microlocal prior for limited-angle tomography (talk)
27th Inverse Days (December 14-16, 2021 in Tampere, Finland), Learning a microlocal prior for limited-angle tomography (talk)
26th Inverse Days (December 14-18, 2020 virtual), 2D Digital Breast Tomosynthesis Reconstruction Using Complex Wavelet Transform (talk)
25th Inverse Days (December 16-18, 2019 in Jyväskylä, Finland), Combining Deep Learning and Iterative Reconstruction in Dental Cone-Beam Computed Tomography (poster)
Enabling Data-Driven Methods for Inverse Problems in 3D Imaging: Uniting Data, People and Algorithms (November 25-29, 2019 in Leiden, Netherlands), Combining Deep Learning and Iterative Reconstruction in Dental Cone-Beam Computed Tomography (poster)
1st Industrial Conference on Artificial Intelligence and Health (November 13-14, 2019 in Milan, Italy), Artificial Intelligence Guided Diagnostics in Medical Imaging (talk)
24th Inverse Days (December 11-13, 2018 in Helsinki, Finland), Denoising of CT projection images using deep learning (poster)

Events and conferences organized:

FCAI-FAME Special Interest Group AI for inverse problems and imaging, kickoff event (March 19, 2025 in Helsinki, Finland)
Finland-Japan Workshop in Industrial and Applied Mathematics (August 26-30, 2024 in Helsinki, Finland)
Finnish Inverse Problems Society career day: Highlighting women in inverse problems (August 7, 2024 in Helsinki, Finland)
AIX Forum: Inverse Problems (November 29, 2023 in Helsinki, Finland)

Public outreach and other talks:

Tutkijoiden yö 2025 (September 6, 2025 in Helsinki, Finland) Käänteisiä ongelmia - varjoista röntgentomografiaan
University of Helsinki (September 15, 2025 in Helsinki, Finland) The mathematics of medical imaging: an introduction to inverse problems and X-ray tomography (lecture)
FAME Connect & Collaborate: Medical Imaging and Technology event (August 28, 2025 in Espoo, Finland) Metal artefact reduction in cone-beam computed tomography
University of Helsinki (August 25, 2025 in Helsinki, Finland) Lääketieteellisen kuvantamisen matematiikkaa: johdatus inversio-ongelmiin ja röntgentomografiaan
MAOLin 90v. liittopäivät (April 26, 2025 in Helsinki, Finland) Tietokoneomografia eli viipalekuvantaminen (interactive workshop)
Tieteen päivät 2025: Nuorten päivä (January 8, 2025 in Helsinki, Finland) Varjojen salaisuus: käänteisten ongelmien kiehtova maailma (interactive workshop)
Mäkelänrinteen lukio (November 5, 2024 in Helsinki, Finland) Röntgentomografiaa ja tekoälyä (lecture)
University of Helsinki (September 23, 2024 in Helsinki, Finland) The mathematics of medical imaging: an introduction to inverse problems and X-ray tomography (lecture)
Kokkolan suomalainen lukio (April 15, 2024 in Kokkola, Finland) Tekoäly (lecture)
Kuvataideakatemia (November 28, 2023 in Helsinki, Finland) Taide tekoälyn aikakaudella (lecture)
MAOLin syysliittokokous (November 11, 2023 in Helsinki, Finland) Tekoäly (talk + workshop)
Hyvinkään lukio (November 9, 2022 in Hyvinkää, Finland) Tekoäly oppii kirjoittamaan (talk)
Kaiken maailman terveys: Värkkää & Törkkää (April 13, 2022 in Helsinki, Finland) Säteilyn matematiikkaa (interactive workshop)
Bring your child to work day (Novermber 22, 2019 in Helsinki, Finland) X-ray Adventure (interactive workshop)
Oulun lukio (October 4, 2019 in Oulu, Finland) Neuroverkot (talk)
Tutkijoiden yö 2019 (September 27, 2019 in Helsinki, Finland) Tietokoneomografia eli viipalekuvantaminen (interactive workshop)
+ Lab visits hosted at the Industrial Mathematics Laboratory at University of Helsinki, Finland

News:

Outside the (black) box: Combining targeted deep learning and inverse mathematics for future potential - FAME FlagshipA Finnish dissertation explored imaging-related inverse problems with machine learning and traditional algorithms. Even old theories can help to break […]

Two weeks of unexpected encounters: Kone and doctoral researchers from the university trialled a new way of developing ideas | University of HelsinkiA new kind of experiment led linguist Aleksandra Dobrego and mathematician Siiri Rautio to a two-week visit to the elevator company. While the expectations for this research sprint were moderate, the outcome of the collaboration turned out to be promising.

New doctoral students starting work on multidisciplinary applications of AI — FCAIEleven new doctoral students at the University of Helsinki are specialising in multidisciplinary applications of artificial intelligence. They were selected from the joint call of the 32 doctoral programmes of the university, in which AI-themed positions were available to applicants from all fields

Public defence: Ill-posed Microlocal Imaging Using Targeted Deep Learning - FAME FlagshipDoctoral Researcher Siiri Rautio will defend her doctoral dissertation Ill-posed Microlocal Imaging Using Targeted Deep Learning in the Faculty of […]

Studia generalia luento tekoälystä Keskiviikkona 17.4. saimme kuunnella Studia generalia luennon tekoälystä. Luennoitsijaksi saimme väitöskirjatutkija Siiri Raution Helsingin...

MATRIX x IMAGINARY: Scaling the Math Outreach Mountain - FAME FlagshipThere are worse places to hold an international conference than the Swiss Alps. Math-related outreach and communications activities offer a […]

Understanding AI: A civic skill for future generations of researchers

Siiri Rautio, a doctoral candidate in mathematics, is developing algorithms for medical imaging in her dissertation project entitled Interpretable Deep Learning for Computed Tomography. The algorithms combine analytical mathematical methods and deep learning.

https://flamma.helsinki.fi/en/group/ajankohtaista/news/-/uutinen/tekoalyn-ymmarrys-on-tulevien-tutkijasukupolvien-kansalaistaito/9274976