{"paper_id":"d2b873e3-268e-42c8-9cdf-c308b8886636","body_text":"2D Contrast-Ultrasound Dispersion Imaging of\nAngiogenesis in Adenomyosis: First Experimental\nMeasurements\nFerenc Igor Kandi\nBiomedical Diagnostics Lab\nEindhoven University of Technology\nEindhoven, the Netherlands\nf.i.kandi@tue.nl\nCatarina Dinis Fernandes\nBiomedical Diagnostics Lab\nEindhoven University of Technology\nEindhoven, the Netherlands\nc.dinis.fernandes@tue.nl\nSimona Turco\nBiomedical Diagnostics Lab\nEindhoven University of Technology\nEindhoven, the Netherlands\ns.turco@tue.nl\nEva de Bock\nGynaecology\nAmsterdam University Medical Center\nAmsterdam, the Netherlands\ne.j.e.debock@amsterdamumc.nl\nLynda Juffermans\nGynaecology\nAmsterdam University Medical Center\nAmsterdam, the Netherlands\nljm.juffermans@amsterdamumc.nl\nJudith Huirne\nGynaecology\nAmsterdam University Medical Center\nAmsterdam, the Netherlands\nj.huirne@amsterdamumc.nl\nMassimo Mischi\nBiomedical Diagnostics Lab\nEindhoven University of Technology\nEindhoven, the Netherlands\nm.mischi@tue.nl\nAbstract—Adenomyosis, a benign yet serious uterine condition,\nis highly prevalent but difficult to diagnose and assess using\nconventional B-mode ultrasound imaging. As the formation of\nadenomyosis involves angiogenesis and influences local microvas-\ncular dispersion, monitoring microvascularity could improve\nadenomyosis diagnostics. Hence, contrast-enhanced ultrasound\n(CEUS), a non-invasive imaging technique, is explored as an\nalternative to B-mode as it allows visualizing microvascularity.\nThe goal is to investigate the potential of characterizing adeno-\nmyosis by quantifying surrogate measures of contrast dispersion,\nusing the contrast-ultrasound dispersion imaging (CUDI) frame-\nwork based on uterine CEUS. CUDI models the passage of an\nultrasound-contrast-agent (UCA) bolus through the uterine tissue\nas a convective dispersion process and expresses UCA dispersion\nkinetics through temporal fitting ( κ) and spatiotemporal similar-\nity analysis. Before applying CUDI on clinical data, the speckle\nsize is regularized to yield isotropy and depth-independence.\nIn addition, the relationship between measured acoustic inten-\nsities and underlying UCA concentrations is determined to be\nlinear ( R2=0.94) for SonoVue T M UCA concentrations of 0.15-\n1.00 mg/L, allowing for the direct application of the proposed\nconvective-dispersion modeling. After regularization, 2D CEUS\nacquisitions of one adenomyotic and one healthy uterus under-\nwent CUDI analysis. The adenomyotic myometrium presented\nheterogeneously higher values of κ and spatiotemporal similarity,\nfor the healthy uterus the values were homogeneous and lower.\nSince κ and spatiotemporal similarity are inversely proportional\nto the level of local dispersion, the results indicate the presence of\nangiogenesis in adenomyotic tissue, supported by the observations\nof heterogeneously decreased local dispersion.\nFunded by NWO TTW, project 18482\nIndex Terms —adenomyosis, uterus, contrast-enhanced ultra-\nsound, contrast-ultrasound dispersion imaging, temporal fitting,\nspatiotemporal similarity\nI. I NTRODUCTION\nBenign uterine conditions, particularly adenomyosis, are\ngreatly prevalent among women [1]. While these conditions\nare classified as benign, they can cause heavy discomfort,\nabnormal bleeding and fertility problems. Prevalence esti-\nmations vary widely, potentially stemming from a current\nlack of robust, objective and quantitative diagnostic methods\n[2]. Current diagnostic protocols rely on B-mode ultrasound\nand magnetic resonance imaging (MRI) to highlight uter-\nine pathologies, but often prove inadequate especially for\nadenomyosis due to its diffuse nature and wide anatomical\nvariability [2]. This highlights the necessity for an objective\ndiagnostic method which can adapt to the widely varying\ncharacteristics of benign uterine conditions, describing them\nboth quantitatively and qualitatively. Adenomyosis has been\nassociated with abnormal vascularization, with angiogenesis\nbeing a particularly important driving mechanism supporting\ndisease progression and development [3]. Since conventional\nB-mode ultrasound does not provide sufficient differential\ndiagnostic power in terms of imaging vascularity, 2D contrast-\nenhanced ultrasound (CEUS) is employed to visualize (mi-\ncro)vasculature and therewith angiogenesis [2]. CEUS is an\nultrasound imaging technique where an ultrasound contrast\nagent (UCA) is intravenously administered and the (mostly)\nThis full-text paper was peer-reviewed at the direction of IEEE Instrumentation and Measurement Society prior to the acceptance and publication.\n2024 IEEE International Symposium on Medical Measurements and Applications (MeMeA) | 979-8-3503-0799-3/24/$31.00 ©2024 IEEE | DOI: 10.1109/MEMEA60663.2024.10596726\nAuthorized licensed use limited to: Universiteit van Amsterdam. Downloaded on June 26,2025 at 11:46:51 UTC from IEEE Xplore.  Restrictions apply. \n\nnonlinear backscatter of the UCA is isolated [4]. Since normal\ntissue backscatter is mostly linear, the UCA signal is differ-\nentiable. In CEUS, adenomyosis has a diffuse appearance,\npresenting both hyper and hypo-enhanced spots.\nContrast-ultrasound dispersion imaging (CUDI) [4] provides\na powerful framework for the quantification of angiogenesis\nbased on CEUS. Within this framework, contrast time-intensity\ncurve fitting and analysis can be used to characterize contrast\ndispersion and flow, based on tracking the wash-in and wash-\nout of a bolus of UCA. In CUDI, the pixel intensities are\ntracked over time and a convective dispersion model is fitted\nto this data. Additionally, by comparing the intensities of a\npixel with its surroundings in time and frequency domain and\nthrough statistical analysis, CUDI can translate this spatiotem-\nporal similarity analysis into dispersion-related metrics [5] [6],\nwhere a decrease in dispersion indicates a more tortuous and\ndense formation of microvascularity, e.g., angiogenesis.\nAdenomyosis spreads heterogeneously throughout the uterine\nmyometrium and angiogenesis takes place to account for the\nlocally increased metabolic demands, decreasing the local\ndispersion. Therefore, the aim is to investigate the use of CUDI\nfor the characterization of adenomyosis, since the hypothesis\nis that the heterogeneity and decreased local dispersion of\nthe adenomyotic vascularity will be reflected in the temporal\nfitting and spatiotemporal similarity analyses.\nII. M ETHODOLOGY\nAlthough CUDI has been extensively used in the prostate\n[5] [7], this is the first time that it is translated to the uterus.\nTherefore, it is required to perform calibrations specific to the\ninvestigated organ and employed ultrasound scanner. For B-\nmode and CEUS image acquisitions, a HERA W10 (Samsung\nMedison, Republic of Korea) ultrasound system with an EV2-\n10A endovaginal transducer is used.\nA. Image preprocessing: speckle size regularization\nSpeckles can be defined as the granular texture on the\nimages, caused by interference of the backscatter signals.\nPrior to any quantitative analysis of the acquired CEUS data,\nspeckle size calibration steps are required, relying on in-vitro\nexperimental measurements.\n1) Rationale: The speckle size is anisotropic throughout\nthe acquisition field of view, which is an effect of both the\ndiverging scanlines (native to using convex transducer arrays)\nand ultrasound propagation physics. As the scan lines diverge,\nthe speckles become increasingly stretched in the lateral\ndirection for increasing depth. For spatiotemporal similarity\nanalysis, CUDI uses an annular kernel and therefore assumes\nan isotropic speckle size. To avoid inefficient computations in\nthe following analyses, i.e., adapting the annular kernel to take\nthis anisotropy into account, the speckle size is regularized to\nensure isotropy in the field of view and the use of one single\nkernel size for the full image [5].\n2) Measurement set-up & protocol: To define the parame-\nters for regularization, the current speckle size distribution is\ndetermined by imaging a known concentration of free-floating\nUCA micro-bubbles using 2D CEUS. The used UCA is\nSonoVueT M (Bracco Suisse SA, Geneva, Switzerland), which\nare micro-bubbles consisting of an inert sulphur hexafluoride\ngas core encapsulated by a phospholipid shell. A large (60-\nL) body of degassed water, contained in an acoustic-padded\npolypropylene plastic container, is used as the imaging vessel.\nThe UCA (SonoVue T M) is prepared directly in the imaging\nvessel to yield the concentration of 0.05 mg/L, as this prede-\ntermined concentration enables measurements of individually\ndiscernible micro-bubbles. The probe is submerged such that\nthe transducer is just below water level and a 20-second\n2D CEUS recording (gain 30 dB, dynamic range 45 dB,\nmechanical index 0.1, 120-degree field of view at 12-cm depth)\nis acquired.\n3) Processing: Working under the assumption that speckles\nhave a Gaussian shape, contour points of the speckles (here-\nwith also the spatial resolution) are estimated though the full-\nwidth at half-maximum (FWHM) of the 2D autocovariance\nfunction (15x15 pixel kernel) [5]. Through these contour\npoints, ellipsoids are fitted using a least-squares approach,\nfollowed by extraction of the axial and lateral axes from\nthe ellipsoids. The dimensions of these axial and lateral axes\ncan then be coupled to the axial and lateral dimensions of\nthe speckles in the unregularized image. The speckle size is\nregularized to be isotropic and equal to the axial size of a\nspeckle in the unregularized image at the depth of interest.\nAssuming a noise-to-signal ratio (NSR) of 1−3 and a σ of\n0.03 cm, Wiener deconvolution is employed to enforce an\nisotropic speckle size [5], followed by low-pass filtering using\nan isotropic Gaussian kernel for high-spatial-frequency noise\nreduction and to obtain a desired resolution [5].\nB. Image preprocessing: concentration-intensity calibration\nFor visualizing backscatter ultrasound waves, the dynamic\nrange is compressed to quantize the signal, typically by using\na logarithm-like function. This compression function therefore\nrelates acoustic intensity to quantized grey level (in this case\nthrough 8-bit greyscale).\n1) Rationale: Having access to this function means that\nUCA concentrations can be assessed in terms of acoustic\nintensity (and not quantization levels), which is assumed\nto be linearly related to lower concentrations of UCA [8],\nsince higher concentrations of UCA will cause self-attenuation\nresulting in lower observed intensities. Establishing and veri-\nfying this relationship is essential for the correct application\nof pharmacokinetic modeling.\n2) Measurement set-up & protocol: The relationship be-\ntween acoustic intensity and UCA concentration is investigated\nby imaging a series of phantoms filled with known (yet\ndifferent) UCA concentrations using 2D CEUS. A large (60-\nL) body of degassed water, contained in an acoustic-padded\npolypropylene plastic container, is again used as the imaging\nvessel. SonoVueT M is prepared to yield the following concen-\ntrations: 0.15; 0.30; 0.50; 1.00; 1.25; 1.80; 2.50; 5.00; 7.50;\n10.00; 12.50; 15.00 and 20.00 mg/L. The rationale behind\nthese concentrations is in that there is no prior knowledge on\nAuthorized licensed use limited to: Universiteit van Amsterdam. Downloaded on June 26,2025 at 11:46:51 UTC from IEEE Xplore.  Restrictions apply. \n\nwhich UCA concentrations can be expected within the uterus,\nhence a wide range of concentrations should be measured\n(focusing on lower concentrations) and taken into account\nwhen relating acoustic intensity and UCA concentration.\nThen, phantoms (condoms) are individually filled with 100 mL\nof each of the prepared UCA concentrations, i.e., one phantom\nper concentration to be measured. The probe is submerged\nsuch that the transducer is just below water level and 20-\nsecond 2D CEUS recordings (gain 30 dB, dynamic range 45\ndB, mechanical index 0.1, 120-degree field of view at 6-cm\ndepth) are acquired for each phantom, moving the phantom\nthrough the field of view while the probe is acquiring (to\naccount for potential concentration inhomogeneities within the\nphantom).\n3) Processing: The means of the measured acoustic inten-\nsities per phantom are calculated and then visualized against\ntheir respective UCA concentrations. A linear least-squared\nerror fit is performed through this data and the goodness of fit\nis assessed through the R2 value. This fit yields the intended\nconcentration-intensity relationship [8].\nC. Clinical data\nThe data of 1 adult female participant with clinically\nconfirmed diffuse adenomyosis and 1 healthy adult female\n(both participants provided written informed consent) was\nacquired within a pilot study of the UteroVue study (NWO\nTTW 18482) at the outpatient gynaecology clinic of the Am-\nsterdam University Medical Center (Amsterdam, the Nether-\nlands). This study has been approved by the Medical Ethics\nCommittee of the Amsterdam University Medical Center\n(NL83391.018.23). Sagittal 2D CEUS recordings of the uteri\nwere performed during peripheral intravenous bolus adminis-\ntrations of SonoVueT M UCA, with the probe fixed in a custom\nprobe holder for stability.\nD. CUDI: time-intensity curve fitting of clinical data\nTo demonstrate the application of the proposed image\noptimization and calibration in clinical data, the uterine 2D\nCEUS recordings are quantitatively analyzed using CUDI. The\nwash-in and wash-out of UCA can be expressed in a time-\nintensity curve (TIC), describing UCA transport kinetics. To\nfit a curve to the acoustic intensities of the backscatter data,\nfirst a preliminary fit to the data is performed with the sole\npurpose of identifying the first pass of the UCA through the\nregion of interest (ROI). Then, a solution to the convective\ndispersion equation in one dimension, a modified local density\nrandom walk model (mLDRW), is used to fit a curve to the\ntime-intensity data based on the first pass. The mLDRW model\ncan be described as [4]\nC(t) = α\ns\nκ\n2π(t − t0) e(− κ(t−t0 −µ)2\n2(t−t0 ) ), (1)\nwhere C(t) is the UCA concentration after an intravenous\nbolus injection as a function of time t, α is a unitless scale\nmetric related to blood flow and UCA dose representing the\narea under the C(t), t0 refers to the theoretical injection\ntime, µ is the mean transit time of a micro-bubble between\nthe injection site and the location of interest. κ is the local,\ndispersion-related parameter governing the shape of C(t)\nrelating flow velocity and dispersion to local microvascular\narchitectures. The value of κ can be expressed as\nκ = v2\n2D [1/s], (2)\nwhere κ relates flow velocity v and dispersion D [8], which\ncannot be measured directly; hence, κ is estimated as a\nsurrogate. The following two cases are evident: lower values\nof κ indicate an increase in dispersion and/or a decrease in\nvelocity, with the TIC being more skewed, while higher values\nof κ indicate a decrease in dispersion and/or an increase in\nvelocity, with the TIC being more symmetric.\nAs described earlier, CUDI analyzes the full wash-in and\nwash-out of a UCA bolus, translating this to local dispersion\ncharacteristics which are expressed through parametric maps\n[8]. In turn, local dispersion can be linked to properties of\nangiogenesis such as microvascular tortuosity and density.\nAlongside κ, CUDI temporal fitting analysis describes local\n(pixel-wise) TICs in terms of appearance time of UCA, mean\ntransit time, time to peak, peak intensity and FWHM (full\ncurve width at half of the maximum peak intensity).\nE. CUDI: spatiotemporal similarity analysis\nSince the spatiotemporal similarity between neighboring\nTICs is governed by local dispersion, spatiotemporal com-\nparison of neighboring TICs yields surrogate estimators of\ndispersion [4]. To perform this comparison, an annular kernel\naround a central pixel is used to avoid a flow-directional bias\n[5]. The dimensions of this kernel are determined to cover\nthe dimensions assumed for early formations of angiogenesis:\nthe inner diameter is determined by the resolution of the\nsystem to be 1 mm, the outer diameter is the angiogenesis-\ninformed value of 2.5 mm. Then, the TIC of the central\npixel is compared to TICs of pixels in the kernel and their\naverage similarity is expressed as temporal correlation, spectral\ncoherence, and mutual information. This process is repeated\nfor all pixels in the ROI and the results are reported in\nthe form of parametric maps. All the mentioned metrics of\nspatiotemporal similarity are inversely proportional to the level\nof local dispersion; a higher similarity indicates less dispersion\nand vice-versa [4].\n1) Temporal correlation: When assessing spatiotemporal\nsimilarity by correlation, the first step is to time-window the\nsignal. Hereafter, the central pixel’s TIC is correlated with\neach TIC in the kernel and these values are averaged to yield\nthe average temporal correlation. This process is repeated for\neach pixel in the ROI to construct the correlation parametric\nmap [5].\n2) Spectral coherence: To avoid time-dependencies due to\nvarying arrival times of the bolus (phase information), spectral\ncoherence regards the TIC signals in frequency (Fourier)\ndomain. The frequency range describing the UCA transport\nkinetics is derived from prostate applications of CUDI [5]\nAuthorized licensed use limited to: Universiteit van Amsterdam. Downloaded on June 26,2025 at 11:46:51 UTC from IEEE Xplore.  Restrictions apply. \n\nand within this range the amplitudes of the frequency spectra\n(no phase) are calculated. From here, the amplitudes of the\nfrequency spectra of the central pixel TIC and the TICs of\npixels in the kernel are correlated and averaged. This process\nis repeated for each pixel in the ROI to construct the coherence\nparametric map [5].\n3) Mutual information: Exploring potential non-linear sim-\nilarities between pixels, mutual information analysis samples\nthe central pixel’s TIC and the TICs of pixels in the kernel\nand regards those sampled signals as random variables. The\nstatistical dependence between those is then measured in terms\nof mutual information. This process is repeated for each pixel\nin the ROI to construct the mutual information parametric map\n[6].\nIII. R ESULTS\nA. Speckle size regularization\nA resulting 2D autocovariance plot can be observed in\nFig. 1, alongside its location as denoted on the CEUS image.\nThe axial and lateral speckle dimensions prior to regularization\ncan be observed in Fig. 2 in blue, where the anisotropy is\nnotable through the consistent sub-millimeter axial size and\nthe increasing lateral size. The determined axial and lateral\ndimensions of the FWHM of the 2D autocovariance can be\nexpressed through:\nF W HMaxial = 0.0631 − 0.0012 × depth, (3)\nF W HMlateral = 0.0539 + 0.116 × depth. (4)\nThe resulting speckle dimensions after regularization can be\nvisually appreciated in Fig. 2 in yellow, where now both the\naxial and the lateral dimensions are consistently around 1 mm\nand the lateral speckle size shows reduced depth-dependency.\nB. Concentration-intensity calibration\nThe linear fit between the averaged acoustic intensity and\nthe lower concentrations (0.15; 0.30; 0.50 & 1.00 mg/L) is\nshown in Fig. 3, reporting an R2 = 0.94. For concentrations\nup to 1.00 mg/L, this linear relationship between UCA con-\ncentrations C [mg/L] and acoustic intensities I [a.u.] can be\nexpressed as\nI = 889.17 × C + 418.46. (5)\nHere, the sensitivity is defined by 889.17 and the background\nacoustic intensity or noise floor, stemming from the surround-\nings, equals 418.46 arbitrary units (a.u.) [8].\nC. CUDI\nAfter applying the proposed regularization and calibration,\nCUDI temporal fitting and spatiotemporal similarity analysis\nis performed on the 2D CEUS acquisitions of the adenomyotic\nuterus and the healthy uterus. Examples of the temporal\nfitting (κ) and spatiotemporal similarity parametric maps for\nthe correlation, coherence and mutual information of the\nmyometrium (muscular tissue layer) of the adenomyotic uterus\nare observable in Fig. 4. Idem, the equivalents for the healthy\nuterus are visualized in Fig. 5. Comparing the uteri, Fig.6\nyields the distributions of the CUDI parameters for the same\nuteri as are observable in Figs. 4-5.\nIV. D ISCUSSIONS\nA. Speckle size regularization\nFrom observing both the CEUS image and the autoco-\nvariance from the denoted region in Fig. 1, it is evident\nthat the speckle sizes are inhomogeneous throughout the\nfield of view, being progressively laterally stretched out with\nincreasing depth. The quantification of this behavior prior\nto any regularization is observable in Fig. 2, demonstrating\nthe increasing lateral speckle size for increasing depth. In\nthis same figure, the effect of regularization is clear through\nthe decrease in depth-dependence of the lateral speckle size\n(the linear fit through the data is more horizontal after reg-\nularization) and the similarity between the axial and lateral\nspeckle sizes (around 1 mm), indicating near isotropy of the\nspeckle dimensions. Important to note is that in the lateral\ndimension, some (although minimal) depth-dependency still\npersists after regularization, therefore showing some room for\nfuture improvement.\nB. Concentration-intensity calibration\nIn the analyzed patients, > 95% of the recorded acoustic\nintensity values are below 1200 a.u., hence calibrating this\nrange takes priority over the higher measured concentrations.\nFrom the measured concentrations in the phantom experiment,\nthis range of acoustic intensities is covered by the lowest\nconcentrations up to 1.00 mg/L, which is observable in Fig. 3.\nHere, the linear fit to the data has an R2 value of 0.94,\nindicating a good fit quality. This indicates that indeed lower\nUCA concentrations can be linearly related to their respective\nmeasured acoustic intensities.\nC. CUDI\nOn standard B-mode ultrasound it proves very difficult to\ndetermine the extent of adenomyosis in the uterus, specifically\nidentifying the microvascularity and boundaries of the affected\ntissue. In contrast to this, when observing the same area in\nCEUS mode, the order of enhancement and the vascularity\n(which is more tortuous, dense and exhibits angiogenesis in\nadenomyotic tissue) provide a significantly clearer picture of\nthe extent of adenomyotic tissue [2]. Yet, the observation\nof the uterus in B-mode and CEUS remains a qualitative\nassessment. Based on CEUS, CUDI proves to be a pow-\nerful framework to quantify the UCA dispersion kinetics\nand translate this to microvascular architecture. Hence, CUDI\ntemporal fitting and spatiotemporal similarity yield quantita-\ntive structural knowledge on the microvasculature and since\nthis directly corresponds to the formation of adenomyosis,\nCUDI is capable of both localizing the adenomyosis-invoked\nangiogenesis and quantifying its extent. As is observable in\nFig. 5, a healthy uterus yields more homogeneous and lower\nvalues for κ and the spatiotemporal similarity metrics, which\ntranslates to higher levels of local dispersion ( κ and the level\nof similarity are both inversely proportional to the level of\nAuthorized licensed use limited to: Universiteit van Amsterdam. Downloaded on June 26,2025 at 11:46:51 UTC from IEEE Xplore.  Restrictions apply. \n\nFig. 1. Local 2D autocovariance (right) for the selected region in the far field of the unregularized CEUS image (left). The speckle size anisotropy can be\neasily observed in the unregularized autocovariance.\nFig. 2. Speckle dimensions before (blue) and after (yellow) regularization,\nincluding axial and lateral linear fits. Observe the more isotropic speckle size\n(comparable axial and lateral dimensions after regularization) and decreased\ndepth-dependence.\nFig. 3. Linear fit ( R2 = 0.94) to the measured mean acoustic intensities of\nthe phantoms with the concentrations 0.15; 0.30; 0.50 and 1.00 mg/L.\nFig. 4. CUDI temporal fitting ( κ) and spatiotemporal similarity parametric\nmaps of the adenomyotic uterus, after implementation of the proposed speckle\nsize regularization. The myometrium is delineated in white on the B-mode\nimage, with therein the endometrium.\nlocal dispersion). In stark contrast to this, an adenomyotic\nuterus demonstrates higher values for κ and the spatiotem-\nporal similarity metrics, and more heterogeneity therein, as\nobservable in Fig. 4. This increase in κ and spatiotemporal\nsimilarity translates to lower levels of local dispersion and\nthe heterogeneity thereof to the diffuse nature of adenomy-\notic anatomy. To quantitatively support these observations,\nFig. 6 demonstrates the increased κ, correlation, coherence\nand mutual information for the adenomyotic myometrium in\ncomparison to the healthy myometrium. Hence, the difference\nin the CUDI quantitative parameters between adenomyosis\nand healthy myometrial tissue is evident from both qualitative\nand quantitative observations of these uteri, but require further\nanalysis in the future on a larger population.\nAuthorized licensed use limited to: Universiteit van Amsterdam. Downloaded on June 26,2025 at 11:46:51 UTC from IEEE Xplore.  Restrictions apply. \n\nFig. 5. CUDI temporal fitting ( κ) and spatiotemporal similarity parametric\nmaps of the healthy uterus, after implementation of the proposed speckle size\nregularization. The myometrium is delineated in white on the B-mode image,\nwith therein the endometrium.\nFig. 6. Distributions of κ, correlation, coherence and mutual information\nfor the aforementioned adenomyotic myometrium (Fig. 4) and the healthy\nmyometrium (Fig. 5). Observe the higher values and larger ranges for the\nadenomyotic tissue in comparison to the healthy tissue.\nV. C ONCLUSIONS\nAdenomyosis is difficult to assess using standard B-mode\nultrasound, hence 2D CEUS was explored as an alternative to\nimage adenomyosis-related formations of angiogenesis and the\ngeneral microvascularity. Prior to quantitative assessment of\nthe CEUS acquisitions, the speckle size was regularized to be\nisotropic and depth-independent throughout the field of view.\nAlso, the relationship between measured acoustic intensities\nof the backscatter signal and lower UCA concentrations was\nfound to be linear. Then, microvascularity was quantitatively\nassessed using CUDI temporal fitting and spatiotemporal\nsimilarity analysis. In adenomyotic tissue, heterogeneously\nhigher κ and spatiotemporal similarity was observed, indicat-\ning decreased local dispersion and an increase in angiogenic\ndevelopment of the microvascularity. Both findings support\nthe hypothesis that the heterogeneity and decreased local\ndispersion of the adenomyotic vascularity will be reflected in\nCUDI analysis.\nACKNOWLEDGMENT\nThis work has been carried out in collaboration with the\nEindhoven University of Technology (Eindhoven, the Nether-\nlands), the Amsterdam University Medical Center (Amster-\ndam, the Netherlands), Bracco Suisse SA (Geneva, Switzer-\nland) and Samsung Medison (Seoul, Republic of Korea).\nREFERENCES\n[1] L. Garcia and K. Isaacson, “Adenomyosis: Review of the Literature,”\nJournal of Minimally Invasive Gynecology, vol. 18, pp. 428—437, 2011.\n[2] B. Stoelinga, L. Juffermans, A. Dooper, M. De Lange, W. Hehenkamp, T.\nVan Den Bosch and J. Huirne, “Contrast-Enhanced Ultrasound Imaging\nof Uterine Disorders: A Systematic Review,” Ultrasonic Imaging, vol.\n43, issue 5, pp. 239—252, 2021.\n[3] M. J. Harmsen, C. F. Wong, V . Mijatovic, A. W. Griffioen, F. Groen-\nman, W. J. Hehenkamp and J. A. Huirne,“Role of angiogenesis in\nadenomyosis-associated abnormal uterine bleeding and subfertility: a\nsystematic review,” Human Reproduction Update, vol. 25, pp. 647–671,\n2011.\n[4] M. Mischi, M. P. J. Kuenen and H. Wijkstra, “Angiogenesis imaging by\nspatiotemporal analysis of ultrasound contrast agent dispersion kinetics,”\nIEEE transactions on ultrasonics, ferroelectrics, and frequency control,\nvol. 59, issue 4, pp. 621–629, 2012.\n[5] M. P. J. Kuenen, T. A. Saidov, H. Wijkstra and M. Mischi, “Contrast-\nUltrasound Dispersion Imaging for Prostate Cancer Localization by\nImproved Spatiotemporal Similarity Analysis,” Ultrasound in Medicine\n& Biology, vol. 39, issue 9, pp. 1631—1641, 2013.\n[6] S. G. Schalk, L. Demi, N. Bouhouch, M. P. J. Kuenen, A. W. Postema,\nJ. J. De La Rosette and M. Mischi, “Contrast-enhanced ultrasound\nangiogenesis imaging by mutual information analysis for prostate cancer\nlocalization,” IEEE Transactions on Biomedical Engineering, vol. 64,\nissue 3, pp. 661—670, 2016.\n[7] C. K. Mannaerts, M. R. Engelbrecht, A. W. Postema, R. A. Van Kollen-\nburg, C. M. Hoeks, C. D. Savci-Heijink and H. Wijkstra, “Detection\nof clinically significant prostate cancer in biopsy-na ¨ıve men: direct\ncomparison of systematic biopsy, multiparametric MRI-and contrast-\nultrasound-dispersion imaging-targeted biopsy,” BJU international, vol.\n126, issue 4, pp. 481—493, 2020.\n[8] M. P. J. Kuenen, M. Mischi and H. Wijkstra, “Contrast-Ultrasound Dif-\nfusion Imaging for Localization of Prostate Cancer,” IEEE Transactions\non Medical Imaging, vol. 30, issue 8, pp. 1493–1502, 2011.\nAuthorized licensed use limited to: Universiteit van Amsterdam. Downloaded on June 26,2025 at 11:46:51 UTC from IEEE Xplore.  Restrictions apply.","source_license":"CC0","license_restricted":false}