著者
加藤 秀起 古賀 佑彦 向山 隆史 戸松 弘孝 鈴木 友輔 鈴木 昇一
出版者
日本保健物理学会
雑誌
保健物理 (ISSN:03676110)
巻号頁・発行日
vol.44, no.4, pp.380-386, 2009 (Released:2011-01-12)
参考文献数
7
被引用文献数
3

In one of the accidents that might happen in the nuclear power station, there is a contamination accident caused by radioactive corrosion products during a periodic inspection. It is necessary to presume the skin absorbed dose from the adhesion area and the contamination density to forecast the level of the skin hazard by the adhesion of the radioactive substance. However, the data to forecast the local skin dose when the radioactive substance adhered handily is not maintained. In this paper, the absorbed dose in the skin surface neighborhood contaminated by radioactive corrosion products was calculated, and the relation between the adhesion area and the contamination density and the local absorption dose was derived. And, the approximate equation that forecast the integrated dose was derived from these data. As for the absorbed dose rate in depth 70 μm from the skin surface that became the index of the skin hazard, the contribution rate by 59Fe was the highest within 30 days, and the contribution of 60Co rose most after the 30th after the radioactive substance had adhered when the contamination density the adhesion area was the same. The relation between the initial contamination density and days that required it was graphed to reaching to the threshold by the integrated dose when the threshold dose in which the necrosis of the skin was caused was assumed to be 20 Gy. The absorbed dose commitment can be presumed from measurements of the contamination density by using this graph or the approximate equation.
著者
小林 正尚 大塚 智子 鈴木 昇一
出版者
公益社団法人 日本放射線技術学会
雑誌
日本放射線技術学会雑誌 (ISSN:03694305)
巻号頁・発行日
vol.69, no.1, pp.19-27, 2013-01-20 (Released:2013-01-25)
参考文献数
18
被引用文献数
9 9

The purpose of this paper is to reappraise the accuracy of a conversion coefficient (k) reported by International Commission on Radiological Protection Publication 102 Table A.2. The effective doses of the routine head computed tomography (CT), the routine chest CT, the perfusion CT, and the coronary CT were evaluated using the conversion coefficient (adult head: 0.021 mSv·mGy-1·cm-1, adult chest: 0.014 mSv·mGy-1·cm-1). The dose length product (DLP) used the value displayed on the console on each scanning condition. The effective doses were evaluated using a human body type phantom (Alderson Rando phantom) and thermoluminescent dosimeter (TLD) elements for comparison with the converted value. This paper reported that the effective doses evaluated from conversion coefficient became different by 0.3 mSv (17%) compared with measurements, the effective dose computed with the conversion coefficient of the adult chest may be underestimated by 45%, and the bolus-tracking which scans the narrow beams should not use a conversion coefficient.
著者
鈴木 昇一
出版者
文教大学
雑誌
情報研究 = Information and Communication Studies (ISSN:03893367)
巻号頁・発行日
vol.24, pp.185-257, 2000-01-01

パターン認識の数学的理論(SS理論)では、入力パターンφに対応するパターンモデルTφを求め、Tφから不動点バターンモデルを連想する形で、φの帰属するカテゴリを決定する多段階パターン変換連想形不動点認識法(SS連想形不動点認識法)が考えられている。本論文では、選ばれた個人の感性を反映するように、"axiom 1を満たすモデル構成作用素"T ,"axiom 2を満たす類似度関数SM", "axiom 3を満たすBSC"の3構造を決める手法が研究されており、SS連想形不動点認識法が感性的パターン情報処理にも適用できることが明らかにされている。\n A muti-stage transformation of patterns has been presented in a mathematical theory(SS theory) of recognizing patterns suggested by S.Suzuki, which gets a corresponding pattern-model Tφ of an input pattern φ in question to be recognized, solves a fixed-point equation of associative reconition about Tφ, and determines a category to which φ belongs so that a fixed-point pattern of a structural fertilization transformation may be recalled by a recognition system RECOGNITRON. A model-construction operator T, a similarity-measure function SM and a rough classifier BSC which must respectively satisfy axiom 1, axiom 2 and axiom 3 are here constructed such that a kansei of a selected person can be reflected in their structures. It became certain that an associative recognition of fixed-point q e suggested by S.Suzuki can be put to a so-called kansei information processing of patterns.
著者
鈴木 昇一
出版者
文教大学
雑誌
情報研究 = Information and Communication Studies (ISSN:03893367)
巻号頁・発行日
vol.19, pp.29-82, 1998-01-01

高次認知機能における推論は、記号列によってなされていると想定すると都合がよい。マルチメディア時代に突入し、マルチメディアの進化形としての知能情報メディアが取り沙汰される現在、記号列とパターンとの2大情報表現を統合する手法の確立が望まれるようになってきた。本研究の目的は、記号列情報処理と同様に精密な推論技術に役立つ1つの論理表現の要素を確保するため、1つの概念はプール関数で符号化されることを勘案し、月本に啓発され、非単調命題論理における論理関数(命題;真理関数)をパターンとみなし、2つの命題の間に或る種の内積を導入し、命題間の距離、命題の持つ情報量を提案している。月本論文では、剰余類の作る線形空間、並びに、線形空間の完備化としてのヒルベルト空間(剰余類の作る空間の完備化)などに言及していないが、本論文では、この不備を補っている。 得られた完全正規直交系を使って、SS理論における3種類のパターンモデルを構成している。このパターンモデルTφは原パターンφの持つ論理構造を簡略化表現しており、モデル構成作用素と呼ばれる作用素Tを使用し、ニューラルネットも構成できる。 本研究によって、命題記号論理をパターン認識分野でのパターンで取り扱うことが可能となった。\n An idea of that an inference in higher-level cognition has been performed using strings has good reason to explain faculties of intelligent information-processing. Two major information-representations by both strings and patterns should be integrated at the present period of intelligent information media. Tsukimoto's paper started me writing this paper. This paper aims at securing a primitive component for logical representation in a higher-level cognitive function in order to serve to make an inference as precisely as symbolic inference. Notice that a concept can be coded with a truth function. In this paper, We regard as a pattern a truth function in non-monotonic propositional logic. A kind of inner product between two propositions is thought out. A distance between two propositions and an amount of information owned by the proposition can be defined Tsukimoto does not make mention of a linear space which consists of residue classes and a Hilbert space as a completion of the linear space. We shall supplement defective descriptions in Tsukimoto's paper. Three kinds of pattern-model proposed by S.Suzuki is constructed using a complete orthonormal system in the obtained Hilbert space. A pattern-model Tφ can represent a corresponding logical structure of the original pattern φ in a simplified form. Moreover It is referred that neural networks are obtained by operator T called model-construction operator. This investigation enables us to treat with a symbolic logic among propositions by means of patterns in the field of pattern-recognition.
著者
鈴木 昇一 佐久間 拓也 前田 英明
出版者
文教大学
雑誌
情報研究 (ISSN:03893367)
巻号頁・発行日
vol.17, pp.133-170, 1996

多段階認識法で用いられる2つのモデル構成作用素T○、T● が構成されている。認識システムは原パターンφを恰も、T○φ、或いは、T●φかのごとく、錯覚して、φの認識処理をすることが可能になる。 得られたモデル構成作用素は、数理形態学における開化作用素◆◆○、閉示作用素◆◆●に対応するものであり、2値化パターンについては、T○=◆◆○、T●=◆◆●と一致するものである。集合の特性関数を2値化パターンφとみなし、T○、◆◆○、T● 、◆◆●のべキ等性などが独自な方法で証明されている。 Two model-construction operators T○, T● are constructed here, which are needed to design a faculties of multi-stage recognition. A recognition system can extract features from Toφ, T●φ instead of φ and therefore can classify Tφ exactly as though Tφ were φ Two operators T○ and T● obtained here correspond to the opening operator ◆◆○ and the closing operator ◆◆● respectively. We shall show that T○φ=◆◆○φ and T●φ=◆◆●φ hold good for any binary pattern φ. The idempotencies of four operators T○, ◆◆○, T●, ●◆◆ are proved using the fact that a binary pattern is equivalent to the corresponding characteristic function of a set.