LoRA-PT:主テンソール単数値とベクトルを用いた海馬のセグメンテーションのためのUNETRを低ランクで適応する
Guanghua He1, Wangang Cheng2, Hancan Zhu2
1Department of Mathematics, Hangzhou Dianzi University, Hangzhou 310018, China; School of Mathematics, Physics and Information, Shaoxing University, Shaoxing 312000, China.
Artificial intelligence in medicine
|September 5, 2025
まとめ
この研究では,ヒポカンプス分割のための新しいパラメータ効率の良い微調整法であるLoRA-PTを導入します. 医療イメージングのデータ不足を解決するために 最低限の更新で高い精度を達成します
科学分野:
- 神経イメージング
- 医療における人工知能
- 計算神経科学
背景:
- 精密な海馬の分割は 精神疾患の理解に不可欠です
- ディープラーニングモデルは高度なセグメンテーションを提供しますが,膨大なリソースとデータが必要です.
- 医療画像のセグメンテーションは,ラベル付きのトレーニングデータが限られているため,しばしば課題に直面します.
研究 の 目的:
- パラメータ効率のよい微調整法 (PEFT) を開発し,海馬のセグメンテーションを行う.
- ヒポカムスのセグメンテーションのために,BraTS2021データセットから,事前に訓練されたUNETRモデルを適応させる.
- 医療イメージングのディープラーニングにおける計算コストとデータ不足の限界を克服する.
主な方法:
- 新しいPEFT技術であるLoRA-PTを提案しました.
- トランスフォーマーパラメータマトリックスから低ランクテンソルを作成するために,テンソル単数値分解を使用した.
- 低ランクテンソールのみを微調整し,残存テンソールを固定します.
主要な成果:
- LoRA-PTは既存のPEFT方法と比較して優れたセグメンテーション精度を示した.
- 微調整に必要なパラメータの更新数を大幅に削減しました.
- 3つの公共の海馬のデータセットで 検証された
結論:
- LoRA-PTは海馬のセグメンテーションに 効率的で効果的な解決策を提供します
- この方法は,大量のデータセットから,限られたデータを持つタスクに,知識を成功裏に転送します.
- LoRA-PTは神経画像分析における ディープラーニングの応用を進めています
関連する概念動画
01:14Base Quantities and Derived Quantities
20.9K
In any system of units, the units for some physical quantities must be specified through a measurement process. These measurements are the base quantities of the system, and their units are the base units of the system. The algebraic combinations of the base values can then be used to express all other physical quantities. Each of these physical quantities is then referred to as a derived quantity, with each unit being referred to as a derived unit.
The International Organization for...
The International Organization for...
20.9K
01:05Estimation of the Physical Quantities
6.7K
On many occasions, physicists, other scientists, and engineers need to make estimates of a particular quantity. These are sometimes referred to as guesstimates, order-of-magnitude approximations, back-of-the-envelope calculations, or Fermi calculations. The physicist Enrico Fermi was famous for his ability to estimate various kinds of data with surprising precision. Estimating does not mean guessing a number or a formula at random. Instead, estimation means using prior experience and sound...
6.7K
01:18Reducing Line Loss
530
In a three-phase circuit, line loss is an indicator of energy dissipated as heat due to the resistance of transmission lines. To address this, incorporating transformers into the system—a step-up transformer at the source and a step-down transformer at the load—is a strategic solution. Two three-phase transformers are introduced to improve this.
With a step-up transformer at the source, the voltage is increased, thereby reducing the current in the transmission lines since power loss...
With a step-up transformer at the source, the voltage is increased, thereby reducing the current in the transmission lines since power loss...
530
01:26Linear Approximation in Frequency Domain
503
Linear systems are characterized by two main properties: superposition and homogeneity. Superposition allows the response to multiple inputs to be the sum of the responses to each individual input. Homogeneity ensures that scaling an input by a scalar results in the response being scaled by the same scalar.
In contrast, nonlinear systems do not inherently possess these properties. However, for small deviations around an operating point, a nonlinear system can often be approximated as linear....
In contrast, nonlinear systems do not inherently possess these properties. However, for small deviations around an operating point, a nonlinear system can often be approximated as linear....
503
01:29Application of Linearization and Approximation
193
A drone flying through complex terrain often relies on more than one sensing method to estimate small changes in altitude. Along with direct measurements, air pressure provides a useful indirect indicator of vertical movement. Atmospheric pressure decreases as altitude increases, and this relationship is commonly described using an exponential model. Although accurate, converting pressure measurements into altitude values requires calculations that are too complex to perform repeatedly during...
193
01:26Trapezoidal Rule
283
Estimating the distance traveled by a vehicle using its recorded velocity over time is a common problem in physics and engineering. When velocity data is available at discrete time intervals, rather than as a continuous function, numerical integration methods such as the trapezoidal rule are often employed to approximate the total displacement.The trapezoidal rule works by dividing the total time interval into several equal segments. Within each segment, the recorded velocities at the endpoints...
283