生物与医学2024science.org
0.05 Tesla 下的全身磁共振成像
Whole-body magnetic resonance imaging at 0.05 Tesla
进阶 · 约 60 分钟读完前置知识：MRI 成像原理、信号处理基础、深度学习基础
导读
这篇论文展示了一种超低场（0.05 特斯拉）全身磁共振成像技术，远低于传统 MRI 的磁场强度，从而大幅降低设备成本和屏蔽要求。研究者在健康志愿者身上采集了脑、脊柱、腹部、肺、肌肉骨骼和心脏图像，并声称深度学习信号预测能有效消除电磁干扰（EMI），无需屏蔽即可获得清晰图像。但评论普遍质疑其清晰图像可能来自神经网络对'正常'影像的生成式补全，而非真实测量信号。
HN 关注点：的核心争议是：用深度学习'清理'低质量医学影像是否会导致算法凭训练集'脑补'出健康图像，从而在诊断上造成危险误导——这与此前三星手机 AI 修月照片引发的争议如出一辙。
HN 评论精选
这几条评论里：质疑 4
质疑AI 增强低质量医学影像会'脑补'正常细节，让放射科医生无法可靠地排除病变，非常危险
几个月前有报道说三星 Galaxy 手机用 AI 对月亮照片做超分，实际是在真实图像质量太差时，根据训练集人为添加地标和细节。文中描述的 AI 增强对放射科医生来说是噩梦：放射科 90% 的工作是在图像质量高时确认'无病变'，质量低时申请补充检查。如果 AI 增强的图像看起来'正常'，医生怎么敢说'我确认没有脑出血'——计算机可能在补偿低画质时错误地补上了'正常'细节。
eig查看英文原文
A few months ago there were articles going around about how Samsung galaxy phones were upscaling images of the Moon using AI [0]. Essentially, the model was artificially adding landmarks and details based on its training set when the real image quality was too poor to make out details. Needless to say, AI upscaling as described in this article would be a nightmare for radiologists. 90% of radiology is confirming the absence of disease when image quality is high, and asking for complementary studies when image quality is low. With AI enhanced images that look "normal", how can the radiologist ever say "I can confirm there is no brain bleed" when the computer might be incorrectly adding "normal" details when compensating for poor image quality? [0] - https://news.ycombinator.com/item?id=35136167
质疑质疑临床医生是否真的接受用 AI 生成的'美化'图像替代真实测量数据
从摘要看，他们是用 ML 合成更高质量、更高分辨率的图像，并以此声称效果可比传统 MRI。临床医生真的宁愿让计算机做'规范化猜测'来美化扫描，也不愿面对反映真实测量的图像、自己运用临床判断吗？
alwa查看英文原文
I can’t access the full paper, but from the abstract, is it accurate that they’re using ML techniques to synthesize higher-quality and higher-resolution imagery, and that’s the basis for their claim that it’s comparable to the output of a conventional MRI scan? Do clinicians really prefer that the computer make normative guesses to “clean up” the scan, versus working with the imagery reflecting the actual measurements and applying their own clinical judgment?
质疑仅用健康志愿者验证远远不够，可能把异常当噪声抹掉；'幻觉出健康'的 MRI 比没有 MRI 更危险
论文说'深度学习信号预测有效消除了 EMI 信号，无需屏蔽即可清晰成像'——那神经网络是不是学会了健康 MRI 长什么样，遇到异常结构时会把它当作 EMI 噪声'纠正'掉，导致误诊？我不否认深度学习在医学影像改进上有重要作用，但这篇论文远远不足以证明这一点。至少需要混合健康/异常患者（包含训练集中不存在的特征），且每个诊断都要在高分辨率机器上复核。必须证明算法不扭曲数据，因为一个'幻觉出健康患者'的 MRI 比没有 MRI 更危险。
cornholio查看英文原文
> We conducted imaging on healthy volunteers, capturing brain, spine, abdomen, lung, musculoskeletal, and cardiac images. Deep learning signal prediction effectively eliminated EMI signals, enabling clear imaging without shielding. So essentially, the neural net was trained to what a healthy MRI looks like and would, when exposed to abnormal structures, correct them away as EMI noise leading to wrong diagnostics? I won't be very dismissive of this approach and probably deep learning has a strong role to play in improving medical imaging. But this paper is far, far from sufficient to prove it. At a minimum, it would require mixed healthy / abnormal patients with particularities that don't exist in the training set, and each diagnostic reconfirmed later on a high resolution machine. You need to actually prove the algorithm does not distort the data, because an MRI that hallucinates a healthy patient is much more dangerous than no MRI at all.
质疑从信号处理角度指出：软件无法无损去除已污染信号的随机噪声，且缺乏体模验证，方法学上站不住脚
这种论文能发表很奇怪。'深度学习消除 EMI'意味着他们找到了去除随机噪声的方法——如果属实这才是真正的革命性声明。但如果 EMI 不是随机的，直接滤波就行；如果是随机的，信号被污染后软件'预测'去除必然损失信息（即分辨率）。他们先用傅里叶变换重建（图像仍很噪），再过神经网络'增强'——到这一步根本不需要信号了，直接告诉网络你想看什么就行。论文没有用已知体模（phantom）做验证扫描，这一点很能说明问题。
elektropionir查看英文原文
It is just weird that papers like this can be published. "Deep learning signal prediction effectively eliminated EMI signals, enabling clear imaging without shielding." - this means that they have found a way to remove random noise, which if true, should be the truly revolutionary claim in this paper. If the "EMI" is not random you can just filter it so you don't need what they are doing. If it isn't random, whatever they are doing can "predict" the noise, they even use the word in that sentence. They are claiming that they can replace physical filtering of noise before it corrupts the signal (shielding) with software "removal" of noise after it has already corrupted the signal. This is simply not possible without loss of information (i.e. resolution). The images that they get from standard Fourier Transform reconstruction are still pretty noisy so on top they "enhance" the reconstruction by running it through a neural net. At that point they don't need the signal - just tell the network what you want to see. The fact that there are no validation scans using known phantoms is telling.
相关资源
从这篇论文所有 HN 讨论帖的全部评论里整理出来的链接。
历次讨论
- 2024Hacker News 讨论帖187 分 · 155 评
同领域还值得读
CRISPR 在活体动物中消除 HIV-1 感染
CRISPR eliminates HIV-1 infection in live animals
报告利用 CRISPR/Cas9 基因编辑技术在活体动物中清除 HIV-1 感染：研究人员通过腺病毒载体递送特定的 Cas9 系统，从被感染细胞中切除嵌入的 HIV 病毒 DNA。
生物与医学2017 · 1,455 分 · 239 评入门 · 约 30 分钟编辑精选保护区飞行昆虫生物量27年下降75%
75% decline over 27 years in total flying insect biomass in protected areas
这篇发表于 PLOS ONE 的研究通过在德国保护区长达 27 年的昆虫生物量监测，发现飞行昆虫总量下降了约 75%。
生物与医学2017 · 901 分 · 432 评入门 · 约 45 分钟编辑精选为什么大多数已发表的研究发现是错的
Why Most Published Research Findings Are False (2005)
这篇 2005 年发表于 PLOS Medicine 的经典论文由 John Ioannidis 撰写，论证了在统计功效不足、研究偏倚和多重检验等因素下，大多数已发表的医学研究发…
生物与医学2022 · 388 分 · 351 评入门 · 约 60 分钟编辑精选人类免疫系统的物理接线图
A physical wiring diagram for the human immune system
这篇 Nature 论文系统绘制了人类免疫系统的“物理接线图”：通过大规模实验鉴定免疫细胞表面受体之间的相互作用（包括约 30 种此前未知的受体互作），并结合定量计算模型模拟免疫…
生物与医学2022 · 342 分 · 49 评专业 · 约 90 分钟编辑精选脂肪组织在减重后仍保留肥胖的表观遗传记忆
Adipose tissue retains an epigenetic memory of obesity after weight loss
这篇 Nature 论文研究了肥胖对脂肪组织的长期影响：即使减重成功，脂肪细胞仍保留着肥胖时期的表观遗传记忆，可能解释为什么减重后容易反弹。
生物与医学2025 · 263 分 · 327 评进阶 · 约 45 分钟编辑精选