WEKO3
アイテム
{"_buckets": {"deposit": "11127650-f6d3-4173-ad62-143b1a67d5cc"}, "_deposit": {"created_by": 1, "id": "70864", "owners": [1], "pid": {"revision_id": 0, "type": "depid", "value": "70864"}, "status": "published"}, "_oai": {"id": "oai:repo.qst.go.jp:00070864", "sets": ["28"]}, "author_link": ["696329", "696327", "696328"], "item_10005_date_7": {"attribute_name": "発表年月日", "attribute_value_mlt": [{"subitem_date_issued_datetime": "2012-09-01", "subitem_date_issued_type": "Issued"}]}, "item_10005_description_5": {"attribute_name": "抄録", "attribute_value_mlt": [{"subitem_description": "Positron emission tomography (PET) enables us to visualize various functional property of living tissue such as neuroreceptor\ndensity. For fully quantitative imaging, time histories of radioactivity concentration in arterial plasma (pTAC)\nand in tissue (tTAC) are measured, and serial arterial blood sampling should be therefore conducted [1]. However, the\nblood sampling expected to be omitted clinically to reduce pain for patients and radiation exposure to medical staffs. Also,\ntTAC is largely contaminated by noise; therefore noise reduction should be applied. A new algorithm named BBD is proposed\nthat estimates noise-free tTAC using a measured noisy tTAC without a measured pTAC based on Bayesian inference. A population\npTAC data-set is utilized as prior information, and physiological feature in a region-of-interest (ROI) is incorporated\ninto BBD to boost the performance.\nAlgorithm: The relationship between tTAC and pTAC is described with a compartment model [1]. Firstly, the following steps\nwere applied to form the prior of noise-free tTAC; 1) a set of artificial pTACs was drawn from a Gaussian distribution, of which\nmean and variance were estimated using a training set of clinically measured population pTACs, 2) a set of physiologically\nfeasible model parameters in the compartment model was defined, and 3) a set of noise-free tTACs was generated for each of\nthe drawn pTACs by utilizing the feasible model parameters. The generated tTACs therefore obeyed the prior distribution of\nnoise-free tTAC. Secondly, given a set of measured tTACs in a ROI, the likelihood for each of the generated noise-free tTACs\nwas computed based on the measurement noise model [2]. The likelihood was multiplied with the prior distribution to obtain the posterior distribution of the noise-free tTAC.\nThirdly, this posterior distribution of the noise free tTAC was then converted to the posterior distribution of pTAC based on the\nfact that tTAC is a function of pTAC. The accuracy of the posterior distribution of pTAC was improved by multiplying all of\nthe posteriors estimated in all ROIs, among which the pTAC was common in the brain. This improved probability distribution\nof pTAC was used to update the prior distribution of tTAC, and the posterior of tTAC was computed again for each measured\ntTAC. The accuracy of the resultant one was improved because of the improvement of the distribution of the pTAC. Finally, a\ndenoised tTAC was derived as an expectation of the posterior\nprobability. Results and discussion: A simulation study was\nconducted to investigate the performance of BBD. sigma1 receptors\nprobe of [11C]SA4503 was selected, and the range of model parameters\nwas taken from [3], and 100 parameters were sampled\nuniformly from the range. Also, 100 clinically measured pTAC\nwere applied to form the pTAC population model. Then, clinically\nmeasured tTACs were simulated, and they were inputter to BBD.\nThe results are presented in the figure. The red curves denote the\ntrue tTAC and the estimations are in blue. The green curves are\nsimulated tTAC. The blue curves were coincided with the red ones,\nand therefore, BBD worked well. The (A) is the result without using ROI information, and (B) is with using ROI. The estimated\nposterior information in gray curves in (B) was narrower than that in (A), that meant ROI information could boost the performance.\nWe can conclude that BBD potentially reduce the noise in PET data dramatically without arterial blood sampling.", "subitem_description_type": "Abstract"}]}, "item_10005_description_6": {"attribute_name": "会議概要(会議名, 開催地, 会期, 主催者等)", "attribute_value_mlt": [{"subitem_description": "34th Annual International Conference of the IEEE Engineering in Medicine \u0026 Biology Society", "subitem_description_type": "Other"}]}, "item_access_right": {"attribute_name": "アクセス権", "attribute_value_mlt": [{"subitem_access_right": "metadata only access", "subitem_access_right_uri": "http://purl.org/coar/access_right/c_14cb"}]}, "item_creator": {"attribute_name": "著者", "attribute_type": "creator", "attribute_value_mlt": [{"creatorNames": [{"creatorName": "Kimura, Yuichi"}], "nameIdentifiers": [{"nameIdentifier": "696327", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "et.al"}], "nameIdentifiers": [{"nameIdentifier": "696328", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "木村 裕一", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "696329", "nameIdentifierScheme": "WEKO"}]}]}, "item_language": {"attribute_name": "言語", "attribute_value_mlt": [{"subitem_language": "eng"}]}, "item_resource_type": {"attribute_name": "資源タイプ", "attribute_value_mlt": [{"resourcetype": "conference object", "resourceuri": "http://purl.org/coar/resource_type/c_c94f"}]}, "item_title": "Baysian Inference Approach for Functional Imaging Using Positron Emission Tomography without Arterial Blood Sampling", "item_titles": {"attribute_name": "タイトル", "attribute_value_mlt": [{"subitem_title": "Baysian Inference Approach for Functional Imaging Using Positron Emission Tomography without Arterial Blood Sampling"}]}, "item_type_id": "10005", "owner": "1", "path": ["28"], "permalink_uri": "https://repo.qst.go.jp/records/70864", "pubdate": {"attribute_name": "公開日", "attribute_value": "2012-09-05"}, "publish_date": "2012-09-05", "publish_status": "0", "recid": "70864", "relation": {}, "relation_version_is_last": true, "title": ["Baysian Inference Approach for Functional Imaging Using Positron Emission Tomography without Arterial Blood Sampling"], "weko_shared_id": -1}
Baysian Inference Approach for Functional Imaging Using Positron Emission Tomography without Arterial Blood Sampling
https://repo.qst.go.jp/records/70864
https://repo.qst.go.jp/records/70864708ae36c-6c93-49e8-97ac-b241228ba8e3
Item type | 会議発表用資料 / Presentation(1) | |||||
---|---|---|---|---|---|---|
公開日 | 2012-09-05 | |||||
タイトル | ||||||
タイトル | Baysian Inference Approach for Functional Imaging Using Positron Emission Tomography without Arterial Blood Sampling | |||||
言語 | ||||||
言語 | eng | |||||
資源タイプ | ||||||
資源タイプ識別子 | http://purl.org/coar/resource_type/c_c94f | |||||
資源タイプ | conference object | |||||
アクセス権 | ||||||
アクセス権 | metadata only access | |||||
アクセス権URI | http://purl.org/coar/access_right/c_14cb | |||||
著者 |
Kimura, Yuichi
× Kimura, Yuichi× et.al× 木村 裕一 |
|||||
抄録 | ||||||
内容記述タイプ | Abstract | |||||
内容記述 | Positron emission tomography (PET) enables us to visualize various functional property of living tissue such as neuroreceptor density. For fully quantitative imaging, time histories of radioactivity concentration in arterial plasma (pTAC) and in tissue (tTAC) are measured, and serial arterial blood sampling should be therefore conducted [1]. However, the blood sampling expected to be omitted clinically to reduce pain for patients and radiation exposure to medical staffs. Also, tTAC is largely contaminated by noise; therefore noise reduction should be applied. A new algorithm named BBD is proposed that estimates noise-free tTAC using a measured noisy tTAC without a measured pTAC based on Bayesian inference. A population pTAC data-set is utilized as prior information, and physiological feature in a region-of-interest (ROI) is incorporated into BBD to boost the performance. Algorithm: The relationship between tTAC and pTAC is described with a compartment model [1]. Firstly, the following steps were applied to form the prior of noise-free tTAC; 1) a set of artificial pTACs was drawn from a Gaussian distribution, of which mean and variance were estimated using a training set of clinically measured population pTACs, 2) a set of physiologically feasible model parameters in the compartment model was defined, and 3) a set of noise-free tTACs was generated for each of the drawn pTACs by utilizing the feasible model parameters. The generated tTACs therefore obeyed the prior distribution of noise-free tTAC. Secondly, given a set of measured tTACs in a ROI, the likelihood for each of the generated noise-free tTACs was computed based on the measurement noise model [2]. The likelihood was multiplied with the prior distribution to obtain the posterior distribution of the noise-free tTAC. Thirdly, this posterior distribution of the noise free tTAC was then converted to the posterior distribution of pTAC based on the fact that tTAC is a function of pTAC. The accuracy of the posterior distribution of pTAC was improved by multiplying all of the posteriors estimated in all ROIs, among which the pTAC was common in the brain. This improved probability distribution of pTAC was used to update the prior distribution of tTAC, and the posterior of tTAC was computed again for each measured tTAC. The accuracy of the resultant one was improved because of the improvement of the distribution of the pTAC. Finally, a denoised tTAC was derived as an expectation of the posterior probability. Results and discussion: A simulation study was conducted to investigate the performance of BBD. sigma1 receptors probe of [11C]SA4503 was selected, and the range of model parameters was taken from [3], and 100 parameters were sampled uniformly from the range. Also, 100 clinically measured pTAC were applied to form the pTAC population model. Then, clinically measured tTACs were simulated, and they were inputter to BBD. The results are presented in the figure. The red curves denote the true tTAC and the estimations are in blue. The green curves are simulated tTAC. The blue curves were coincided with the red ones, and therefore, BBD worked well. The (A) is the result without using ROI information, and (B) is with using ROI. The estimated posterior information in gray curves in (B) was narrower than that in (A), that meant ROI information could boost the performance. We can conclude that BBD potentially reduce the noise in PET data dramatically without arterial blood sampling. |
|||||
会議概要(会議名, 開催地, 会期, 主催者等) | ||||||
内容記述タイプ | Other | |||||
内容記述 | 34th Annual International Conference of the IEEE Engineering in Medicine & Biology Society | |||||
発表年月日 | ||||||
日付 | 2012-09-01 | |||||
日付タイプ | Issued |