Strategies for Improving Detection of Magnetic Nanoparticles in Magnetomotive Ultrasound
(2026) BMEM01 20261Division for Biomedical Engineering
- Abstract
- Magnetomotive ultrasound (MMUS) is a developing imaging modality using magnetic nanoparticles (MNPs) as contrast agent. MNPs are accumulated at a tissue target and set in motion by a time-varying magnetic field. The resulting displacement spreads to surrounding tissue and is detected with ultrasound. The objective of this modality is to transition medical ultrasound from structural to molecular specific imaging, with application potential for cancer cell detection and drug administration. An existing MNP detection method is based on quadrature demodulation and phase gating. However, the detection limit remains constrained by the method’s sensitivity to noise. Additionally, cardiorespiratory rhythms increase the risk of spectral leakage in... (More)
- Magnetomotive ultrasound (MMUS) is a developing imaging modality using magnetic nanoparticles (MNPs) as contrast agent. MNPs are accumulated at a tissue target and set in motion by a time-varying magnetic field. The resulting displacement spreads to surrounding tissue and is detected with ultrasound. The objective of this modality is to transition medical ultrasound from structural to molecular specific imaging, with application potential for cancer cell detection and drug administration. An existing MNP detection method is based on quadrature demodulation and phase gating. However, the detection limit remains constrained by the method’s sensitivity to noise. Additionally, cardiorespiratory rhythms increase the risk of spectral leakage in the detection of fixed frequency magnetic excitation commonly used for MMUS. To overcome these limitations, I implemented three strategies to improve the detection of MNPs in MMUS and evaluated their efficiency with a contrast-to-noise ratio (CNR) on tissue-mimicking phantoms with descending MNP concentrations. The first strategy was frequency modulated magnetic excitation signals, specifically linear bidirectional chirp signals generated with a rotating permanent magnet. Secondly, I developed a matched filter algorithm that correlates phase oscillation in ultrasound radio frequency (RF) data with a magnetic force reference. The resulting normalized correlation was employed as a probability mask, where each pixel value represents the likelihood of MNP presence, used for weighting the results of the previous detection method. The third strategy was implementation of thresholding for segmentation and morphological filtering to suppress noise and refine object boundaries on the mask. The chirp excitation signals did not substantially improve the visibility of accumulated MNP regions compared to fixed frequency excitation signals. Thus, the use of these signals is not justified. When the previous detection method was combined with the probability mask from the matched filter, the CNR increased by 6.6 dB for 0.5 mg/ml of MNPs, 7.6 dB for 0.3 mg/ml of MNPs and 12.6 dB for 0.1 mg/ml of MNPs. Segmentation and morphology further improved the CNR with 1.6 dB for 0.5 mg/ml of MNPs, 1.7 dB for 0.3 mg/ml of MNPs and 16 dB for 0.1 mg/ml of MNPs. These results demonstrate that the matched filter algorithm and image processing enhance noise suppression and spatial specificity in MMUS, likely allowing detection of lower MNP concentrations. (Less)
- Popular Abstract (Swedish)
- Så hittas cancer med magneter och ultraljud
Genom att kombinera ultraljud med magnetiskt kontrastmedel kan man hitta cancerceller som tidigare varit osynliga. Detta banar vägen för billigare cancerdiagnostik utan strålning. En algoritm som utvecklats i detta arbete gör metoden ännu mer pricksäker.
risk vävnad och tidiga cellförändringar ser nästan identiska ut med vanligt ultraljud. Att försöka skilja dem åt är därför som att leta efter en nål i en höstack. Jag har utvecklat en algoritm för magnetomotoriskt ultraljud som kan måla upp gränsen mellan friskt och sjukt med högre noggrannhet än tidigare metoder. Denna teknik öppnar dörren för cancerdiagnostik som är strålningsfri, billigare och mer tillgänglig än dagens metoder.
Idag... (More) - Så hittas cancer med magneter och ultraljud
Genom att kombinera ultraljud med magnetiskt kontrastmedel kan man hitta cancerceller som tidigare varit osynliga. Detta banar vägen för billigare cancerdiagnostik utan strålning. En algoritm som utvecklats i detta arbete gör metoden ännu mer pricksäker.
risk vävnad och tidiga cellförändringar ser nästan identiska ut med vanligt ultraljud. Att försöka skilja dem åt är därför som att leta efter en nål i en höstack. Jag har utvecklat en algoritm för magnetomotoriskt ultraljud som kan måla upp gränsen mellan friskt och sjukt med högre noggrannhet än tidigare metoder. Denna teknik öppnar dörren för cancerdiagnostik som är strålningsfri, billigare och mer tillgänglig än dagens metoder.
Idag används ofta dyra tekniker som MR och PET för att hitta specifika cancerceller. Ultraljud hade varit ett perfekt komplement till dessa om det inte vore för den dåliga kontrasten. Lösningen kan vara magnetiska järnpartiklar som kontrastmedel. Dessa små partiklar injiceras i kroppen och söker upp cancerceller. Genom att sedan skaka om dem med ett starkt magnetfält får man omgivande vävnad att vibrera. Det är dessa vävnadsvibrationer som upptäcks med ultraljudet i tekniken som kallas för magnetomotoriskt ultraljud.
Min nya algoritm fungerar som ett avancerat filter. Genom att matcha vävnadens rörelse med magnetfältets svängningar kan algoritmen rensa bort brus och fokusera helt på vibrationerna från järnpartiklarna. Det gör att även mycket små mängder kontrastmedel syns tydligt. Detta är viktigt för att tekniken ska bli så känslig som möjligt för diagnostik.
För att testa algoritmen gjordes experiment på vävnadsliknande fantomer, tänk som ett bröstimplantat med en konstgjord tumör i. En roterande magnet skapade magnetfältet som fick järnpartiklarna i ''tumören'' att vibrera samtidigt som en ultraljudsgivare tog bilder. Hypotesen var att om magneten snurrade med varierande rotationsfrekvenser skulle vibrationerna vara lättare att hitta. Resultatet visade oväntat nog att det fungerar lika bra med en konstant frekvens. Däremot ger den nya algoritmen noggrannare detektion, bara genom att analysera samma data på ett annat sätt. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/student-papers/record/9234890
- author
- Stickel, Sophia LU
- supervisor
-
- Tomas Jansson LU
- Maria Evertsson LU
- Jules Reniaud LU
- organization
- alternative title
- Strategier för att förbättra detektion av magnetiska nanopartiklar i magnetomotoriskt ultraljud
- course
- BMEM01 20261
- year
- 2026
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Magnetomotive Ultrasound, Contrast Agent, Matched Filter, Contrast-to-Noise Ratio, Segmentation, Morphology
- language
- English
- additional info
- 2026-09
- id
- 9234890
- date added to LUP
- 2026-06-29 09:14:55
- date last changed
- 2026-06-29 09:14:55
@misc{9234890,
abstract = {{Magnetomotive ultrasound (MMUS) is a developing imaging modality using magnetic nanoparticles (MNPs) as contrast agent. MNPs are accumulated at a tissue target and set in motion by a time-varying magnetic field. The resulting displacement spreads to surrounding tissue and is detected with ultrasound. The objective of this modality is to transition medical ultrasound from structural to molecular specific imaging, with application potential for cancer cell detection and drug administration. An existing MNP detection method is based on quadrature demodulation and phase gating. However, the detection limit remains constrained by the method’s sensitivity to noise. Additionally, cardiorespiratory rhythms increase the risk of spectral leakage in the detection of fixed frequency magnetic excitation commonly used for MMUS. To overcome these limitations, I implemented three strategies to improve the detection of MNPs in MMUS and evaluated their efficiency with a contrast-to-noise ratio (CNR) on tissue-mimicking phantoms with descending MNP concentrations. The first strategy was frequency modulated magnetic excitation signals, specifically linear bidirectional chirp signals generated with a rotating permanent magnet. Secondly, I developed a matched filter algorithm that correlates phase oscillation in ultrasound radio frequency (RF) data with a magnetic force reference. The resulting normalized correlation was employed as a probability mask, where each pixel value represents the likelihood of MNP presence, used for weighting the results of the previous detection method. The third strategy was implementation of thresholding for segmentation and morphological filtering to suppress noise and refine object boundaries on the mask. The chirp excitation signals did not substantially improve the visibility of accumulated MNP regions compared to fixed frequency excitation signals. Thus, the use of these signals is not justified. When the previous detection method was combined with the probability mask from the matched filter, the CNR increased by 6.6 dB for 0.5 mg/ml of MNPs, 7.6 dB for 0.3 mg/ml of MNPs and 12.6 dB for 0.1 mg/ml of MNPs. Segmentation and morphology further improved the CNR with 1.6 dB for 0.5 mg/ml of MNPs, 1.7 dB for 0.3 mg/ml of MNPs and 16 dB for 0.1 mg/ml of MNPs. These results demonstrate that the matched filter algorithm and image processing enhance noise suppression and spatial specificity in MMUS, likely allowing detection of lower MNP concentrations.}},
author = {{Stickel, Sophia}},
language = {{eng}},
note = {{Student Paper}},
title = {{Strategies for Improving Detection of Magnetic Nanoparticles in Magnetomotive Ultrasound}},
year = {{2026}},
}