DC6 PhD in Mathematical Imaging: Denoising and Material Decomposition for Photon-Counting CT

DC6 PhD in Mathematical Imaging: Denoising and Material Decomposition for Photon-Counting CT

Research cut-off: 6 November 2026
Assessment: Promising with material uncertainties

Quick Decision

DC6 is a mathematically focused doctoral project at the interface of University of Innsbruck (UIBK) mathematics and Medical University of Innsbruck (MUI) radiology. Its central work is to combine denoising with material decomposition in photon-counting CT (PCCT), using mathematical modelling and machine learning and developing efficient, task-adapted diffusion-model strategies for spectral CT data. The current official application page gives 10 December 2026 as the first-call deadline, and the official projects page lists DC6. This conflicts with the supplied position information’s 1 March 2027 deadline field and expired marker; use the current official call as the best evidence at the 6 November cut-off, then recheck the portal. Apply · DC6 project

The strongest fit is someone with substantial, demonstrable mathematics and Python/programming experience; direct PCCT experience is not required by the project description. Local research ties mathematical reconstruction and radiology AI to PCCT, and a recent local study has examined PCCT iodine maps. Still, public sources do not promise that the DC6 researcher will use the newly installed PCCT scanner, a particular dataset or clinical workflow. The key administrative conflicts are the listed deadline and a supplied 10 December 2026 start field versus the call’s expected early-February 2027 employment start. Confirm the controlling dates, employer, research resources and doctoral route before relying on them.

Research Problem and Doctoral Work

What Is Unsolved and Why It Matters

Photon-counting detectors sort X-ray events into energy ranges, enabling spectrally resolved CT and material-specific information. The project description frames a joint problem: reduce noise while decomposing spectral images into material information, using mathematical modelling and machine learning. The challenge is that material estimates and image noise interact; a method that improves image appearance may distort material maps, while a decomposition can amplify uncertainty in lower-count channels. The proposed work is to develop efficient learning strategies adapted to this problem, with diffusion models among the methods named for spectral CT. Potential applications include multispectral CT and automated stroke prediction using PCCT. These are research aims and application contexts, not established DC6 diagnostic systems or validated clinical performance. DC6 project description

Work, Methods, and Potential Research Angles

The advertised tasks are mathematical modelling, machine-learning method development, and implementation/evaluation of algorithms that jointly address denoising and material decomposition. The position information highlights inverse problems, optimization, Python and algorithm assessment. A useful proposal could clarify how a diffusion prior is conditioned on spectral measurements, how image quality and quantitative material-map accuracy should be evaluated together, and how learned methods should be tested when the spectral channels, scanner settings or material distributions differ from training data. These are evidence-informed questions rather than promised deliverables.

Recent work by the local mathematics group presents derivative-free iterative reconstruction for multispectral CT. A 2026 external study compares diffusion-posterior-sampling strategies for PCCT material decomposition and explores hybrid regularization when materials are absent from training. A separate 2023 study on preclinical photon-counting micro-CT denoising emphasizes generalization across energy thresholds. Together, these examples suggest a field balancing reconstruction fidelity, noise, quantitative material accuracy, computational efficiency and generalization. The preclinical study and external PCCT results do not guarantee DC6 outcomes. Local multispectral CT paper · PCCT diffusion study · PCCT denoising study

Recent Team and Field Evidence

• Prohaszka, Neumann and Haltmeier (2024) proposed a derivative-free iterative one-step method for multispectral CT reconstruction. It is closely related mathematical inverse-problem work from UIBK, but it does not study DC6’s specific PCCT diffusion pipeline. Paper
• Tiefenthaler et al. (2025) studied shape-aware extraction of head/neck arteries from CT angiography images. This is team-linked evidence of CT/AI method work, not PCCT material decomposition. Paper
• Pali et al. (2026) examined how PCCT iodine-map signal relates to visual detectability on conventional brain-window images. The retrospective study included 12 cases and describes its thresholds as preliminary and in need of larger independent validation. It demonstrates local PCCT activity, not an automated stroke-prediction system or a DC6 deliverable. Paper
• Vazia et al. (2026) compared diffusion posterior sampling for PCCT material decomposition and tested a hybrid with variational regularizers for materials not included in training. This is direct external evidence for the project’s method area; it cannot be assumed to work on DC6’s data. Paper
• Nuyts et al. (2023) evaluated a denoising network on preclinical photon-counting micro-CT across energy thresholds and material maps. Its scanner and data differ from clinical PCCT; it is useful as a generalization question, not a performance benchmark for DC6. Paper

The MUI radiology research report states that a PCCT scanner was installed in 2025 and that PCCT would become a research focus. This is relevant institutional context, but it does not confirm DC6 scanner access, patient data or compute allocation. Radiology research report

School, Programme and Field Context

DC6 names Lukas Neumann of UIBK and Stephanie Mangesius of MUI as supervisors. The official UIBK Engineering Mathematics pages identify Neumann with mathematical/medical imaging and the Image-guided Diagnosis and Therapy research area. The MUI AI Methods in Radiology group lists Mangesius among its associated researchers and Neumann as a cooperation partner; the group’s remit includes CT, dual-energy CT, PCCT, quantitative image analysis and AI. This is a useful collaboration signal, but it does not define the division of supervision or guarantee a particular clinical/technical facility. UIBK research · AI Methods in Radiology

The advertised degree route is the UIBK Doctoral Programme in Engineering Sciences: full time, six semesters/180 ECTS, award Dr. techn., with German or English listed as study languages. Admission generally requires a relevant master’s/diploma or equivalent and a suitable dissertation project; the general doctoral admission page sets out registration documentation. Clarify whether employment is legally with UIBK or MUI and which doctoral office handles enrolment. IGNIDTE describes core/elective classes, journal clubs, thematic seminars, training in medical imaging/AI/MRI/reconstruction, Python and statistics, and a career-development plan. The project information describes two international secondments; programme materials say partner arrangements are still to be announced, so destinations and timing remain unconfirmed. Engineering Sciences curriculum · Doctoral admission · Training

QS lists UIBK at #333 in the 2027 overall World University Rankings. This is not a mathematics, radiology or PCCT subject ranking; UIBK cautions that ranking methods vary. The university is located in Innsbruck, Austria. QS profile · UIBK ranking guidance · Location

Supervisor and Lab Assessment

Mathematical and Radiology Environment

Neumann’s UIBK group works on mathematical modelling and methods for biomedical/engineering applications and is associated with IGDT; the local radiology AI group includes Mangesius and cooperates with Neumann. The research report describes multiple hospital imaging resources and a PCCT scanner installed in 2025. Recent local papers cover multispectral reconstruction, CT/CTA inference and PCCT iodine maps, which collectively support scientific activity near DC6. Public materials do not specify DC6’s mathematical supervision split, scanner access, data permissions, computational resources or dedicated research budget. Ask which datasets are approved, how much of the project is mathematical theory versus empirical algorithm development, and what support is available for evaluating domain shift.

Funding, Application, and Living Cost

Contract and Date Conflicts

The current official call and UIBK’s scientific-staff pay scale state a three-year, 40-hour-per-week doctoral employment contract at €3,776.10 gross per month, 14 times per year under the 2026 rate, plus possible allowances and benefits. Net pay is to be communicated at interview. The supplied position information lists an offer-start date of 10 December 2026, while the official call expects employment after offers and admission in early February 2027. Request the contractual start date and payroll terms in writing. First-call announcement · UIBK salary information

The official application page gives 10 December 2026 as the deadline, while the supplied record’s structured deadline field says 1 March 2027 at 23:59 Europe/Vienna. The current project description also says 10 December. At the 6 November research cut-off, rely on the current official call and recheck the portal. The position information asks applicants to select DC6 and submit a two-page research proposal; the official call evaluates academic qualifications, research, project compatibility, proposal, English/communication and complementary skills. Interviews are expected online 18–31 January 2027 and employment in early February. Applicants must disclose any generative-AI use in application documents; non-English documents need certified English translations where applicable, and no application fee is required. Application page

Eligibility and Living Costs

Formal eligibility requires no prior doctorate, a relevant master’s or equivalent degree by employment start, and no more than 12 months living, working or studying in Austria during the 36 months before 10 December 2026; permanent host employees are excluded. The European Commission and REA explain the general MSCA mobility framework, while the current call controls the recruitment. EC MSCA Financial Guide · REA eligibility

UIBK gives an approximate student budget of €1,200/month (€450 accommodation, €350 food and €400 study/recreation). OeAD’s national student estimate is about €1,300/month, including accommodation €450–700, food around €300 and other study/personal costs around €500. Neither is an employee budget. Net salary, fees/ÖH contributions, housing cost and any relocation support are not established by these estimates. UIBK stay guidance · OeAD living costs

Career Market Context

AMS lists 4,921 data-scientist postings in Austria in 2025 and a generally positive 2026–2028 outlook for that broad occupation. Statistics Austria reported 123,400 vacancies across sectors in Q2 2026. These are broad labour-market indicators, not forecasts for PCCT, medical-imaging reconstruction or diffusion-model research roles. Adjacent pathways could include applied mathematics, medical imaging, CT reconstruction, data science, radiology AI and research engineering, but these are possibilities rather than employment guarantees. AMS data-scientist barometer · Statistics Austria vacancies

Applicant Fit and Preparation

Required Background and Skills

The formal eligibility conditions are the degree, no-prior-doctorate and MSCA mobility/host-employment rules. Separately, the DC6 position seeks a master’s-level applicant with strong mathematics and Python coding skills, interested in inverse problems, modelling, machine learning, optimization, and algorithm development/evaluation. The ideal profiles include mathematics with coding, physics or engineering with documented mathematical study, or computer science with a strong mathematical focus. Direct experience in PCCT is not required; purely application-oriented training without evidenced mathematical depth is described as a poor fit.

Preferred Experience and Suggested Preparation

Before applying, assemble transcript/course evidence for analysis, linear algebra, probability or stochastic processes and examples of Python-based numerical or machine-learning work. A two-page proposal could make explicit the target material maps, the role of noise, how model-based and learned approaches will be compared, and how generalization will be assessed. These are suggestions for demonstrating fit; the call’s formal screens remain degree and mobility eligibility.

Next Actions and Questions to Verify

Before Applying

• Recheck that DC6 is available and confirm that 10 December 2026 is the controlling deadline, including exact local time and time zone.
• Verify the 12-in-36-month mobility calculation and prepare final-degree/expected-completion evidence.
• Check the full online upload list, certified translation/legalization requirements, two-page proposal format and AI-use disclosure.
• Clarify how the preferred-project selection should be entered and whether the proposal can refine the tentative topic.

At Interview

• Ask which institution is the employer and which unit manages the doctoral registration.
• Confirm whether the start date is early February 2027 and how the input’s 10 December 2026 start field arose.
• Ask about mathematical scope, the balance between modelling and implementation, access to PCCT scanner/data, compute and clinical collaborators, evaluation criteria, and supervisor roles.
• Confirm the two secondments’ partner selection, timing and cost coverage.

Before Accepting an Offer

• Obtain written salary, net-pay estimate, allowances, fees/ÖH obligations and contract dates.
• Confirm scanner/data access and the support/resources tied to the work plan.
• Confirm doctoral registration, supplementary coursework if any, and secondment terms.

Evidence and Sources

Official call and project

• IGNIDTE Apply
• IGNIDTE DC6 project page
• IGNIDTE Training
• IGNIDTE first-call announcement

School, supervisors and environment

• UIBK Doctoral Programme Engineering Sciences
• UIBK doctoral admission
• UIBK ranking guidance
• QS World University Rankings 2027
• UIBK student cost guidance
• OeAD living costs
• UIBK locations
• UIBK Engineering Mathematics research
• Lukas Neumann — UIBK department team
• AI Methods in Radiology
• MUI Radiology research report

Research papers and methods

• Prohaszka et al. (2024), multispectral CT reconstruction
• Tiefenthaler et al. (2025), shape-aware CTA inference
• Tiefenthaler et al. (2025), unsupervised CTA calcification learning
• Pali et al. (2026), PCCT iodine-map visibility
• Vazia et al. (2026), diffusion models for PCCT material decomposition
• Nuyts et al. (2023), PCCT micro-CT denoising
• External multi-material PCCT decomposition (2024)
• Rybertt et al. (2026), clinical PCCT K-edge feasibility

Funding, mobility and living costs

• UIBK scientific-staff salary scales
• European Commission MSCA Financial Guide
• REA MSCA eligibility
• UIBK locations

Career market

• Statistics Austria job-vacancy data
• AMS JobBarometer: data scientists